<?xml version="1.0" encoding="utf-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9"
        xmlns:image="http://www.google.com/schemas/sitemap-image/1.1">
  <url>
    <loc>https://data-today.net/inference-price-uneven/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/inference-uneven.png</image:loc>
      <image:title>Inference got cheap, but not for the work that pays</image:title>
      <image:caption>Inference got cheap, but not for the work that pays</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/compute-stock-doubling/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/compute-stock.png</image:loc>
      <image:title>The world&#39;s AI compute now doubles every seven months</image:title>
      <image:caption>The world&#39;s AI compute now doubles every seven months</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/context-windows-30x/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/context-windows.png</image:loc>
      <image:title>Context windows grew 30x a year. Your retrieval stack noticed</image:title>
      <image:caption>Context windows grew 30x a year. Your retrieval stack noticed</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/consumer-hardware-lag/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/consumer-hardware.png</image:loc>
      <image:title>Last year&#39;s frontier model now runs on a laptop</image:title>
      <image:caption>Last year&#39;s frontier model now runs on a laptop</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/gigawatt-buildout/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/datacenter-buildout.png</image:loc>
      <image:title>The gigawatt build-out is the real AI race now</image:title>
      <image:caption>The gigawatt build-out is the real AI race now</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/open-models-default/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/open-models-hero.png</image:loc>
      <image:title>Open-weight models closed the gap to 1.7 points</image:title>
      <image:caption>Open-weight models closed the gap to 1.7 points</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/agentic-cancellation-cliff/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/agentic-cancellation.png</image:loc>
      <image:title>Most agentic AI projects will be cancelled before they ship</image:title>
      <image:caption>Most agentic AI projects will be cancelled before they ship</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/training-cost-vs-efficiency/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cost-vs-efficiency.png</image:loc>
      <image:title>Training costs rise 3.5x a year. Efficiency is the only brake</image:title>
      <image:caption>Training costs rise 3.5x a year. Efficiency is the only brake</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/scaling-curve-extrapolate/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/compute-scatter.png</image:loc>
      <image:title>The scaling curve nobody wants to extrapolate</image:title>
      <image:caption>The scaling curve nobody wants to extrapolate</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/compute-geography/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/compute-geography.png</image:loc>
      <image:title>Three-quarters of the world&#39;s AI compute sits in one country</image:title>
      <image:caption>Three-quarters of the world&#39;s AI compute sits in one country</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/everyone-automated-everything/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/automation-hero.png</image:loc>
      <image:title>Everyone automated everything. Then they hired more people.</image:title>
      <image:caption>Everyone automated everything. Then they hired more people.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/benchmark-heatmap-explainer/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/benchmark-heatmap.png</image:loc>
      <image:title>The model leaderboard is tightening to a photo finish</image:title>
      <image:caption>The model leaderboard is tightening to a photo finish</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/hardware-value-per-dollar/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/hardware-value.png</image:loc>
      <image:title>AI chips give 24x more per dollar, if you can afford the sticker</image:title>
      <image:caption>AI chips give 24x more per dollar, if you can afford the sticker</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/agents-working-too-hard/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/agent-throughput-line.png</image:loc>
      <image:title>Your agents are working too hard</image:title>
      <image:caption>Your agents are working too hard</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/capability-acceleration/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/capability-accel.png</image:loc>
      <image:title>AI capability stopped slowing down. It sped up after 2024</image:title>
      <image:caption>AI capability stopped slowing down. It sped up after 2024</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/labs-promise-agi-consultants/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/model-costs-chart.png</image:loc>
      <image:title>The bill for frontier AI is now measured in gigawatts</image:title>
      <image:caption>The bill for frontier AI is now measured in gigawatts</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/vibecoding-to-production/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/vibecoding-hero.png</image:loc>
      <image:title>Vibe coding is loud online and rare in real codebases</image:title>
      <image:caption>Vibe coding is loud online and rare in real codebases</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/debate-data-cleaning/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/debate-data-cleaning.png</image:loc>
      <image:title>Multi-agent debate needs a boring data-cleaning cop</image:title>
      <image:caption>Multi-agent debate needs a boring data-cleaning cop</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/google-ai-opt-out-lever/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/google-ai-opt-out-lever.png</image:loc>
      <image:title>Google AI opt out gives publishers a real lever now</image:title>
      <image:caption>Google AI opt out gives publishers a real lever now</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/unit-distance-openai-proof/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/unit-distance-openai-proof.png</image:loc>
      <image:title>Unit distance proof moves AI past clever math demos</image:title>
      <image:caption>Unit distance proof moves AI past clever math demos</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/audit-then-score-ground-truth/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/audit-then-score-ground-truth.png</image:loc>
      <image:title>Audit then score turns AI ground truth into a process</image:title>
      <image:caption>Audit then score turns AI ground truth into a process</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/generalist-agents-data-curation/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/generalist-agents-data-curation.png</image:loc>
      <image:title>Generalist agents hit data curation’s hard budget wall</image:title>
      <image:caption>Generalist agents hit data curation’s hard budget wall</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/virtual-power-plants-data-centers/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/virtual-power-plants-data-centers-figure.png</image:loc>
      <image:title>Virtual power plants meet AI’s data center power wall</image:title>
      <image:caption>Horizontal bar chart on a log scale comparing three capacity figures in megawatts.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/ai-energy-water-per-query/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ai-energy-water-per-query-fig-energy.png</image:loc>
      <image:title>What an AI prompt really costs in energy, water and CO2</image:title>
      <image:caption>Bar chart of energy per AI request.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/ai-persuasion-reddit-field-test/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ai-persuasion-reddit-field-test-fig-persuasive-rate.png</image:loc>
      <image:title>AI persuasion got its Reddit field test, and it was ugly</image:title>
      <image:caption>Horizontal bar chart of Reddit AI persuasion delta rates.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/eva-bench-voice-agents/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/eva-bench-voice-agents.png</image:loc>
      <image:title>EVA-Bench Data 2.0 makes voice agents measurable</image:title>
      <image:caption>EVA-Bench Data 2.0 makes voice agents measurable</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/meta-agent-security-fail/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/meta-agent-security-fail-fig-mythos-benchmark-gap.png</image:loc>
      <image:title>AI agent security failed in Meta’s support desk test</image:title>
      <image:caption>Grouped bar chart for AI agent security context.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/palantir-data-software-worth/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/palantir-data-software-worth-fig-revenue.png</image:loc>
      <image:title>Why Palantir&#39;s data software is worth $400 billion</image:title>
      <image:caption>Area chart showing Palantir revenue rising from 1.1 billion dollars in 2020 to 1.5 billion.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/teampcp-threat-actor/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/teampcp-threat-actor-fig.png</image:loc>
      <image:title>TeamPCP: the crew behind the npm worm hitting Red Hat</image:title>
      <image:caption>Horizontal bar chart on a log scale comparing repositories involved in three TeamPCP waves.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/world-cup-2026-dynamic-pricing/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/world-cup-2026-dynamic-pricing-fig.png</image:loc>
      <image:title>World Cup 2026 tickets: dynamic pricing takes over</image:title>
      <image:caption>Bar chart comparing the most expensive regular final ticket.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/covert-llm-persuasion-audits/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/covert-llm-persuasion-audits.png</image:loc>
      <image:title>Covert LLM agents need persuasion audits, not labels</image:title>
      <image:caption>Covert LLM agents need persuasion audits, not labels</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/llm-judges-follow-up/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/llm-judges-follow-up-fig-judge-cost-gap.png</image:loc>
      <image:title>LLM judges crack under post-decision challenge tests</image:title>
      <image:caption>Horizontal bar chart showing human annotation at $300 per 1,000 examples.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/sentinelbench-monitoring-agents/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/sentinelbench-monitoring-agents-fig-sentinelstep-reliability.png</image:loc>
      <image:title>SentinelBench tests whether AI agents can wait well</image:title>
      <image:caption>Bar chart comparing monitoring task success rates.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/ai-data-center-backlash/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ai-data-center-backlash-fig-data-center-power-curve.png</image:loc>
      <image:title>AI data center backlash hits Shelbyville&#39;s $2B bet</image:title>
      <image:caption>Line chart for AI data center backlash showing U.S.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/databricks/databricks-data-intelligence-platform/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/databricks-data-intelligence-platform.png</image:loc>
      <image:title>Databricks Data Intelligence Platform, explained for builders</image:title>
      <image:caption>Databricks Data Intelligence Platform, explained for builders</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/matillion/matillion-context-engine/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/matillion-context-engine.png</image:loc>
      <image:title>Matillion Context Engine grounds Maia agent work</image:title>
      <image:caption>Matillion Context Engine grounds Maia agent work</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/matillion/matillion-data-productivity-cloud/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/matillion-data-productivity-cloud.png</image:loc>
      <image:title>Matillion&#39;s Data Productivity Cloud, explained for builders</image:title>
      <image:caption>Matillion&#39;s Data Productivity Cloud, explained for builders</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/password-cracking-heatmap/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/password-cracking-heatmap.png</image:loc>
      <image:title>How long would it take to brute force your password with today&#39;s computing power?</image:title>
      <image:caption>How long would it take to brute force your password with today&#39;s computing power?</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/snowflake-cortex-agents/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/snowflake-cortex-agents.png</image:loc>
      <image:title>Building an AI agent on your Snowflake data with Cortex Agents</image:title>
      <image:caption>Building an AI agent on your Snowflake data with Cortex Agents</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/snowflake-cortex-ai-functions/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/snowflake-cortex-ai-functions-fig.png</image:loc>
      <image:title>Snowflake Cortex: running LLMs where your data already lives</image:title>
      <image:caption>Bar chart of relative Cortex cost per million tokens by model tier.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/snowflake-cortex-analyst/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/snowflake-cortex-analyst.png</image:loc>
      <image:title>Cortex Analyst: let business users query Snowflake in plain English</image:title>
      <image:caption>Cortex Analyst: let business users query Snowflake in plain English</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/snowflake-cortex-search/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/snowflake-cortex-search.png</image:loc>
      <image:title>Cortex Search: the managed RAG engine inside Snowflake</image:title>
      <image:caption>Cortex Search: the managed RAG engine inside Snowflake</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/snowflake-gen2-warehouses/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/snowflake-gen2-warehouses.png</image:loc>
      <image:title>Snowflake Gen2 warehouses: faster compute you have to opt into</image:title>
      <image:caption>Snowflake Gen2 warehouses: faster compute you have to opt into</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/snowflake-governance-rbac/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/snowflake-governance-rbac.png</image:loc>
      <image:title>Snowflake RBAC and masking: lock it down without grinding to a halt</image:title>
      <image:caption>Snowflake RBAC and masking: lock it down without grinding to a halt</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/snowflake-iceberg-tables/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/snowflake-iceberg-tables.png</image:loc>
      <image:title>Should you put your data lake in Snowflake Iceberg tables?</image:title>
      <image:caption>Should you put your data lake in Snowflake Iceberg tables?</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/snowflake-openflow/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/snowflake-openflow.png</image:loc>
      <image:title>Snowflake Openflow: managed ingestion built on Apache NiFi</image:title>
      <image:caption>Snowflake Openflow: managed ingestion built on Apache NiFi</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/snowflake-performance-clustering/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/snowflake-performance-clustering.png</image:loc>
      <image:title>Snowflake clustering: when a clustering key actually pays off</image:title>
      <image:caption>Snowflake clustering: when a clustering key actually pays off</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/snowflake-pipelines-streams-tasks/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/snowflake-pipelines-streams-tasks-fig.png</image:loc>
      <image:title>Dynamic Tables vs Streams and Tasks: which Snowflake pipeline to build</image:title>
      <image:caption>Horizontal bar chart of lowest practical lag by Snowflake approach.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/snowflake-warehouse-sizing/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/snowflake-warehouse-sizing-fig-credits.png</image:loc>
      <image:title>Snowflake warehouse sizing: stop paying for idle compute</image:title>
      <image:caption>Bar chart of Snowflake credits per hour by warehouse size.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/attack-selection-agent-safety/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/attack-selection-agent-safety-fig-audit-budget-drop.png</image:loc>
      <image:title>Attack selection makes AI agent safety look too high</image:title>
      <image:caption>Horizontal bar chart showing attack selection reduces measured safety at a 1% audit budget.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/crowdmath-math-agents-gap/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/crowdmath-math-agents-gap-fig-reasoning-gap.png</image:loc>
      <image:title>CrowdMath exposes the math gap AI agents still miss</image:title>
      <image:caption>CrowdMath results show next post prediction accuracy at 88 percent and post role classification.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/london-robotaxi-waitlist/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/london-robotaxi-waitlist-fig-london-robotaxi-trust.png</image:loc>
      <image:title>London robotaxi race starts with Uber interest list</image:title>
      <image:caption>Donut chart showing London robotaxi trust.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/snowflake-iceberg-merge-read/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/snowflake-iceberg-merge-read.png</image:loc>
      <image:title>ICEBERG_MERGE_ON_READ_BEHAVIOR changes Iceberg DML</image:title>
      <image:caption>ICEBERG_MERGE_ON_READ_BEHAVIOR changes Iceberg DML</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/ai-plateau-goalpost/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ai-plateau-goalpost-fig-saturation.png</image:loc>
      <image:title>The AI plateau is mostly a goalpost that keeps moving</image:title>
      <image:caption>Line chart of top-model scores by year.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/fable-5-mythos-5-routing/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/fable-5-mythos-5-routing-fig-fallback.png</image:loc>
      <image:title>Fable 5 and Mythos 5 are one model with a bouncer</image:title>
      <image:caption>Donut chart titled where a Fable 5 call actually runs.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/miasma-worm-agent-repos/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/miasma-worm-agent-repos-fig-microsoft-repo-distribution.png</image:loc>
      <image:title>Miasma worm turns AI coding agents into repo traps</image:title>
      <image:caption>Horizontal bar chart of Miasma worm disabled Microsoft GitHub repos.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/seattle-data-center-moratorium/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/seattle-data-center-moratorium-fig-global-demand-doubles.png</image:loc>
      <image:title>Seattle data center moratorium tests AI’s power grab</image:title>
      <image:caption>Line chart showing global data center electricity use rising from 415 TWh in 2024.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/xcode-apple-ai-platform-toll/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/xcode-apple-ai-platform-toll-fig-afm-preference-gap.png</image:loc>
      <image:title>Xcode 27 makes Apple AI the next app platform toll</image:title>
      <image:caption>Bar chart showing AFM 3 Core preferred on 45.6 percent of general text prompts.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/confidential-inference-apple-cloud/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/confidential-inference-apple-cloud-fig-afm-preference-gains.png</image:loc>
      <image:title>Confidential inference makes Apple’s cloud AI real</image:title>
      <image:caption>Bar chart of Apple human preference results.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/mcp-explained/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/mcp-explained-fig-math.png</image:loc>
      <image:title>What is MCP? The Model Context Protocol for data engineers</image:title>
      <image:caption>Line chart comparing connector counts: without MCP.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/siri-ai-google-rollout/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/siri-ai-google-rollout-fig-model-preference-gains.png</image:loc>
      <image:title>Siri AI makes Apple’s platform bet depend on Google</image:title>
      <image:caption>Bar chart of Siri AI model evaluations.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/whatsapp-ai-gatekeeping-test/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/whatsapp-ai-gatekeeping-test-fig-fine-ceiling.png</image:loc>
      <image:title>WhatsApp AI assistants become EU gatekeeping test</image:title>
      <image:caption>Donut chart for WhatsApp AI assistants enforcement showing a maximum fine ceiling of $20.1 billion.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/agentic-commerce-visa-chatgpt/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/agentic-commerce-visa-chatgpt-fig-merchant-readiness.png</image:loc>
      <image:title>Agentic commerce gets Visa’s ChatGPT payment rails</image:title>
      <image:caption>Bar chart of agentic commerce merchant readiness in 2026.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/fable-guardrails-visibility-reset/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/fable-guardrails-visibility-reset-fig-guardrail-trigger-rates.png</image:loc>
      <image:title>Claude Fable 5 guardrails get a visibility reset</image:title>
      <image:caption>Claude Fable 5 guardrails chart showing visible fallback at fewer than 5 percent of sessions.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/multi-agent-safety-google-funding/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/multi-agent-safety-google-funding-fig-agent-success-gap.png</image:loc>
      <image:title>Multi-agent safety gets Google’s $10 million test</image:title>
      <image:caption>Donut chart showing OSWorld agent outcomes in 2025.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/amazon-data-center-water/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/amazon-data-center-water-fig-wue-comparison.png</image:loc>
      <image:title>Amazon data centers put 2.5B gallons on the AI bill</image:title>
      <image:caption>Amazon data centers used 0.12 L/kWh in 2025.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/generative-engine-optimization/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/generative-engine-optimization-fig-tactics.png</image:loc>
      <image:title>Generative engine optimization: how to rank inside an LLM</image:title>
      <image:caption>Bar chart of GEO tactics and their effect on source visibility in AI answers.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/miasma-hack-live/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
  </url>
  <url>
    <loc>https://data-today.net/vscode-autopilot-default-risk/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/vscode-autopilot-default-risk-fig-developer-environment-share.png</image:loc>
      <image:title>VS Code Autopilot puts agent risk in every default</image:title>
      <image:caption>Horizontal bar chart showing Visual Studio Code at 75.9 percent usage.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/agentperf-blackwell-agent-benchmark/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/agentperf-blackwell-agent-benchmark-fig-blackwell-relative-lift.png</image:loc>
      <image:title>AgentPerf puts Blackwell’s agent lead at 20x per watt</image:title>
      <image:caption>Horizontal bar chart of published NVIDIA benchmark lifts vs H200 baseline.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/olmo-eval-model-loop/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/olmo-eval-model-loop-fig-mmlu-scoring-gap.png</image:loc>
      <image:title>olmo-eval brings statistical discipline to LLM loops</image:title>
      <image:caption>MMLU macro scores from OLMES: OLMo-7B scored 28.3 MCF macro and 40.5 CF macro.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/snowflake-hybrid-tables-speed/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/snowflake-hybrid-tables-speed-fig-hybrid-table-gains.png</image:loc>
      <image:title>Snowflake Hybrid Tables get faster without request fees</image:title>
      <image:caption>Bar chart of Snowflake Hybrid Tables gains showing point lookup throughput at 8x.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/snowflake-iceberg-adls-writes/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/snowflake-iceberg-adls-writes-fig-polling-cost-knobs.png</image:loc>
      <image:title>Snowflake Iceberg ADLS writes are finally usable</image:title>
      <image:caption>Snowflake Iceberg ADLS operating chart showing 30 seconds default polling.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/workbench-agents-reliability-gap/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/workbench-agents-reliability-gap-fig-workbench-outcomes.png</image:loc>
      <image:title>WorkBench agents close the workplace reliability gap</image:title>
      <image:caption>Stacked chart of WorkBench agents outcomes across 690 tasks.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/eu-ai-content-labelling/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/eu-ai-content-labelling-fig-deadline-compression.png</image:loc>
      <image:title>AI content labelling gives builders 47 days to move</image:title>
      <image:caption>Line chart for AI content labelling showing 270 days left at the November 5.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/flexible-data-centers-grid-dial/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/flexible-data-centers-grid-dial-fig-grid-headroom.png</image:loc>
      <image:title>Flexible data centers turn AI power into a grid dial</image:title>
      <image:caption>Line chart for flexible data centers and other large loads showing US grid headroom.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/searchleak-copilot-trust-boundary/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/searchleak-copilot-trust-boundary-fig-copilot-cvss-scores.png</image:loc>
      <image:title>SearchLeak turns Copilot’s trust boundary into the bug</image:title>
      <image:caption>Horizontal bar chart showing SearchLeak Microsoft CNA score 6.5.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/divinit-agentic-search-threads/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/divinit-agentic-search-threads-fig-qwen3-divinit-scale.png</image:loc>
      <image:title>DivInit makes agentic search threads less wasteful</image:title>
      <image:caption>Line chart for DivInit agentic search showing Qwen3 multi-hop QA pass@4.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/llm-recommendation-bias-brands/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/llm-recommendation-bias-brands.png</image:loc>
      <image:title>LLM recommendation bias gives famous brands an edge</image:title>
      <image:caption>LLM recommendation bias gives famous brands an edge</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/mlperf-blackwell-scale-edge/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/mlperf-blackwell-scale-edge-fig-scale-points.png</image:loc>
      <image:title>MLPerf Training 6.0 gives Blackwell the scale edge</image:title>
      <image:caption>MLPerf Training 6.0 scale chart showing 8,192 GPUs for DeepSeek-V3 on Blackwell.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/ai-trust-gap-chatbots/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ai-trust-gap-chatbots-fig-chatgpt-adoption-curve.png</image:loc>
      <image:title>AI trust gap widens as chatbot use hits 49% of US adults</image:title>
      <image:caption>Line chart for AI trust gap context showing ChatGPT use among U.S.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/vmware-migration-broadcom-risk/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/vmware-migration-broadcom-risk-fig-vmware-pressure.png</image:loc>
      <image:title>VMware migration turns Broadcom risk into a roadmap</image:title>
      <image:caption>Horizontal bar chart of VMware migration pressure.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/large-load-ferc-test/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/large-load-ferc-test-fig-data-center-demand.png</image:loc>
      <image:title>Large-load interconnection gets a 60-day FERC test</image:title>
      <image:caption>Line chart showing US data center electricity use rising from 58 TWh in 2014.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/mosaicleaks-agent-query-secrets/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/mosaicleaks-agent-query-secrets-fig-success-leakage-tradeoff.png</image:loc>
      <image:title>MosaicLeaks shows research agents leak query secrets</image:title>
      <image:caption>Scatter chart of MosaicLeaks Qwen3-4B results.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/subquadratic-attention-hard-test/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/subquadratic-attention-hard-test-fig-prefill-latency.png</image:loc>
      <image:title>Subquadratic attention gets its first serious test</image:title>
      <image:caption>Line chart comparing SubQ SSA and FlashAttention-2 prefill latency.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/agentic-clinical-rag-acceptance/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/agentic-clinical-rag-acceptance-fig-acceptance-range.png</image:loc>
      <image:title>Agentic clinical RAG gets 96.5% clinician acceptance</image:title>
      <image:caption>Bar chart for agentic clinical RAG showing 80 percent lowest per-type clinician acceptance.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-mastra-npm-supply-chain/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-mastra-npm-supply-chain.png</image:loc>
      <image:title>Mastra npm supply chain attack hits AI build rooms</image:title>
      <image:caption>Mastra npm supply chain attack hits AI build rooms</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/diffusion-language-models-benchmark-tax/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/diffusion-language-models-benchmark-tax-fig-reported-speed.png</image:loc>
      <image:title>Diffusion language models meet a messy benchmark tax</image:title>
      <image:caption>Horizontal bar chart of reported diffusion language model speeds.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/ram-price-shock-nothing-phone/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ram-price-shock-nothing-phone-fig-memory-index.png</image:loc>
      <image:title>RAM price shock just killed Nothing’s budget phone</image:title>
      <image:caption>Line chart of the RAM price shock showing conventional DRAM at 100 in Q4 2025.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/ai-music-training-data/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ai-music-training-data-fig-music-dataset-scale.png</image:loc>
      <image:title>AI music training data gets a searchable paper trail</image:title>
      <image:caption>AI music training data chart comparing four reported collections.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-fortibleed-fortigate-credentials/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-fortibleed-fortigate-credentials-fig-fortibleed-exposure-counts.png</image:loc>
      <image:title>FortiBleed puts FortiGate credentials on the clock now</image:title>
      <image:caption>Bar chart for FortiBleed exposure showing CISA at about 74,000 Fortinet devices and SOCRadar.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-splunk-cve-siem-clock/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-splunk-cve-siem-clock-fig-splunk-cve-timeline.png</image:loc>
      <image:title>Splunk CVE-2026-20253 puts SIEMs on a 72-hour clock</image:title>
      <image:caption>Bar chart for Splunk CVE-2026-20253 showing disclosure at day 0 on June 10.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/databricks/databricks-lakeflow-designer-cost/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/databricks-lakeflow-designer-cost-fig-designer-surface-area.png</image:loc>
      <image:title>Databricks Lakeflow Designer GA cost guide for teams</image:title>
      <image:caption>Databricks Lakeflow Designer exposes 9 built-in AI Function operators.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/databricks/databricks-pat-auto-scoping/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/databricks-pat-auto-scoping-fig-pat-scoping-windows.png</image:loc>
      <image:title>Databricks PAT auto-scoping is a quiet job breaker</image:title>
      <image:caption>Horizontal bar chart for Databricks PAT auto-scoping showing 7 days for the reminder before enforcement.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/matillion/matillion-anomaly-alerts/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/matillion-anomaly-alerts-fig-baseline-window.png</image:loc>
      <image:title>Matillion anomaly alerts: runtime drift without polling</image:title>
      <image:caption>Bar chart for Matillion anomaly alerts showing 10 successful runs required before alerts and up.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/snowflake-adaptive-compute-ga/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/snowflake-adaptive-compute-ga-fig-price-performance-index.png</image:loc>
      <image:title>Snowflake Adaptive Compute changes AWS cost models</image:title>
      <image:caption>Bar chart showing Gen2 standard warehouses at a 1.0 price-performance index and Snowflake Adaptive Compute.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/snowflake-dynamic-tables-refresh/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/snowflake-dynamic-tables-refresh-fig-change-detection-checks.png</image:loc>
      <image:title>Snowflake Dynamic Tables refresh gets a cost test</image:title>
      <image:caption>Bar chart for Snowflake Dynamic Tables showing 200 tables at 1 minute target lag produce.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/vibe-coding-publish-gate/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/vibe-coding-publish-gate-fig-public-assets-risk.png</image:loc>
      <image:title>Vibe coding security needs a real publish gate now</image:title>
      <image:caption>Donut chart for vibe coding security showing 5,000 sensitive-data assets out of 380,000 public AI-built.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/copilot-spend-meter/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/copilot-spend-meter-fig-individual-credit-ladder.png</image:loc>
      <image:title>Copilot spend meter shows where AI coding costs bite</image:title>
      <image:caption>Bar chart of individual Copilot AI credit allowances.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-autojack-agent-rce/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-autojack-agent-rce-fig-autogen-public-footprint.png</image:loc>
      <image:title>AutoJack makes AI agent prototypes a real RCE risk</image:title>
      <image:caption>AutoGen public footprint chart showing 59.2k GitHub stars and 8.9k GitHub forks for the Microsoft.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/high-na-euv-bottleneck/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/high-na-euv-bottleneck-fig-high-na-resolution.png</image:loc>
      <image:title>High-NA EUV turns AI chips into a $400M bottleneck</image:title>
      <image:caption>Line chart comparing ASML EUV platforms.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/nvidia-hotter-water-defense/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/nvidia-hotter-water-defense-fig-water-drop.png</image:loc>
      <image:title>45C liquid cooling gives Nvidia a tougher water defense</image:title>
      <image:caption>Step chart for 45C liquid cooling showing conventional cooling-tower systems at 2,600,000 gallons per megawatt.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/oracle-ai-layoffs-cloud-capital/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/oracle-ai-layoffs-cloud-capital-fig-oracle-headcount-drop.png</image:loc>
      <image:title>Oracle AI layoffs expose the cloud capital tradeoff</image:title>
      <image:caption>Line chart for Oracle AI layoffs showing Oracle full-time employees falling from 162,000 in fiscal.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/pqc-deadline-2030-clock/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/pqc-deadline-2030-clock.png</image:loc>
      <image:title>Post-quantum cryptography deadline gets a 2030 clock</image:title>
      <image:caption>Post-quantum cryptography deadline gets a 2030 clock</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/rift-bench-agent-red-teaming/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/rift-bench-agent-red-teaming-fig-domain-asr.png</image:loc>
      <image:title>RIFT-Bench tests agents where prompts cannot reach</image:title>
      <image:caption>RIFT-Bench deterministic attack success rates by domain.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/agentic-ai-work-delegation/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/agentic-ai-work-delegation-fig-codex-task-horizons.png</image:loc>
      <image:title>Agentic AI work shifts from chat to delegation at scale</image:title>
      <image:caption>Agentic AI work task-horizon chart showing 80.6 percent of sampled individual Codex users exceeded 30.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-cisa-kev-edge-gear/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-cisa-kev-edge-gear.png</image:loc>
      <image:title>CISA KEV vulnerabilities put edge gear on watch now</image:title>
      <image:caption>CISA KEV vulnerabilities put edge gear on watch now</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/ford-ai-quality-veterans/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ford-ai-quality-veterans-fig-iqs-rebound.png</image:loc>
      <image:title>Ford AI quality fix puts veterans back in the loop</image:title>
      <image:caption>Line chart of JD Power initial quality scores showing overall PP100 falling from 192.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/ibm-nanostack-chip-moores-law/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ibm-nanostack-chip-moores-law-fig-density-doubling.png</image:loc>
      <image:title>IBM nanostack chip gives Moore’s Law a vertical path</image:title>
      <image:caption>Line chart for IBM nanostack chip density milestones.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/benchmark-saturation-agent-signal/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/benchmark-saturation-agent-signal-fig-failure-roots.png</image:loc>
      <image:title>Benchmark saturation still leaves AI agents exposed</image:title>
      <image:caption>Bar chart for benchmark saturation failure roots showing wrong metric or computation at 18 failures.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/coding-agent-rewards-verification/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/coding-agent-rewards-verification-fig-monitoring-shift.png</image:loc>
      <image:title>Coding agent rewards hit a harder verification horizon</image:title>
      <image:caption>Bar chart for coding agent rewards showing clean resolved at 40.22 percent before monitoring.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/gpt-56-delay-permit-race/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/gpt-56-delay-permit-race-fig-mythos-triage-funnel.png</image:loc>
      <image:title>GPT-5.6 delay makes AI model launches a permit race</image:title>
      <image:caption>Horizontal bar chart for GPT-5.6 delay context showing Anthropic&#39;s Mythos vulnerability triage.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-cisco-unified-cm-deadline/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-cisco-unified-cm-deadline-fig-cisco-unified-cm-patch-clock.png</image:loc>
      <image:title>Cisco Unified CM CVE gets a weekend patch clock</image:title>
      <image:caption>Bar chart for Cisco Unified CM CVE-2026-20230 showing 22 days from Cisco advisory to KEV.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-signal-backup-keys/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-signal-backup-keys-fig-messaging-warning-timeline.png</image:loc>
      <image:title>Signal backup recovery keys become the weak link for ops</image:title>
      <image:caption>Line chart showing Signal Secure Backups announced at 0 days.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/llm-novel-search-space/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/llm-novel-search-space-fig-bands.png</image:loc>
      <image:title>How to make an LLM escape its own prior with geometry</image:title>
      <image:caption>A 2D KNN-style map.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/mythos-access-whitelist/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/mythos-access-whitelist-fig-approved-access-map.png</image:loc>
      <image:title>Mythos 5 access returns as a government whitelist</image:title>
      <image:caption>Horizontal bar chart showing Fable 5 public access at 0 organizations.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/apple-ai-price-hike/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/apple-ai-price-hike-fig-apple-hardware-price-hikes.png</image:loc>
      <image:title>Apple AI price hike makes memory a consumer tax now</image:title>
      <image:caption>Horizontal bar chart of Apple hardware price increases.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/apple-cxmt-memory-risk/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/apple-cxmt-memory-risk-fig-memory-price-jump.png</image:loc>
      <image:title>Apple CXMT memory deal makes RAM a policy fight now</image:title>
      <image:caption>Line chart for the Apple CXMT memory deal showing conventional DRAM price increase midpoints.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/ai-jobs-transition-europe/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ai-jobs-transition-europe-fig-eu-us-transition-mix.png</image:loc>
      <image:title>AI jobs transition puts Europe in redesign mode now</image:title>
      <image:caption>Stacked chart comparing AI jobs transition shares.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/ai-peer-review-scale-test/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ai-peer-review-scale-test-fig-submission-surge.png</image:loc>
      <image:title>AI peer review gets a serious scale test at ICML</image:title>
      <image:caption>Line chart showing combined ICLR.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-ai-coding-agent-malware/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-ai-coding-agent-malware.png</image:loc>
      <image:title>AI coding agent malware hides in clean GitHub repos</image:title>
      <image:caption>AI coding agent malware hides in clean GitHub repos</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-vscode-tasks-supply-chain/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-vscode-tasks-supply-chain-fig-affected-packages-by-ecosystem.png</image:loc>
      <image:title>VS Code Tasks supply chain attack needs new checks</image:title>
      <image:caption>Bar chart for the VS Code Tasks supply chain attack showing 2 affected npm packages.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/databricks/databricks-ai-gateway-budgets/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/databricks-ai-gateway-budgets-fig-genie-thresholds.png</image:loc>
      <image:title>Unity AI Gateway budgets: the spend guardrail guide</image:title>
      <image:caption>Unity AI Gateway budgets example for Genie showing $5,000 shared threshold.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/lineshine-cpu-scale-politics/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/lineshine-cpu-scale-politics-fig-top-five-hpl.png</image:loc>
      <image:title>LineShine supercomputer makes CPU scale political again</image:title>
      <image:caption>Horizontal bar chart showing LineShine leading the June 2026 TOP500 at 2.198 exaflops.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/matillion/matillion-maia-bigquery-support/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/matillion-maia-bigquery-support-fig-bigquery-component-coverage.png</image:loc>
      <image:title>Matillion Maia BigQuery support: setup, cost, limits</image:title>
      <image:caption>Bar chart for Matillion Maia BigQuery support showing 29 transformation components.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/snowflake-dynamic-iceberg-replication/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/snowflake-dynamic-iceberg-replication-fig-replication-usage-windows.png</image:loc>
      <image:title>Dynamic Iceberg table replication, costed honestly</image:title>
      <image:caption>Replication usage history retention is 14 days in Information Schema table functions and 365 days.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/ai-music-royalties-tidal-line/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ai-music-royalties-tidal-line-fig-deezer-ai-upload-surge.png</image:loc>
      <image:title>AI music royalties face Tidal’s harder line in 2026</image:title>
      <image:caption>Line chart of Deezer fully AI-generated music uploads.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/claude-gb300-azure-agents/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/claude-gb300-azure-agents-fig-gb300-vs-gb200.png</image:loc>
      <image:title>Claude on GB300 puts Azure agents on serious new iron</image:title>
      <image:caption>Horizontal bar chart showing GB300 versus GB200.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/spring-ai-java-agent-stack/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/spring-ai-java-agent-stack.png</image:loc>
      <image:title>Spring AI 2.0 gives Java agents a sturdier app stack</image:title>
      <image:caption>Spring AI 2.0 gives Java agents a sturdier app stack</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/ai-browser-security-bioshocking/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ai-browser-security-bioshocking.png</image:loc>
      <image:title>AI browser security fails the six-agent guardrail test</image:title>
      <image:caption>AI browser security fails the six-agent guardrail test</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/claude-science-pharma-plumbing/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/claude-science-pharma-plumbing-fig-science-resource-scale.png</image:loc>
      <image:title>Claude Science turns lab agents into pharma plumbing</image:title>
      <image:caption>Horizontal bar chart showing selected Claude Science resources on a log scale.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-langflow-rce-miner-endpoints/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-langflow-rce-miner-endpoints-fig-operator-clock.png</image:loc>
      <image:title>Langflow RCE turns AI app endpoints into miner bait</image:title>
      <image:caption>Langflow RCE chart with two bars.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-simplehelp-cve-patch-drill/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-simplehelp-cve-patch-drill-fig-simplehelp-exposure.png</image:loc>
      <image:title>SimpleHelp CVE-2026-48558 needs a patch drill</image:title>
      <image:caption>Bar chart for SimpleHelp CVE-2026-48558 showing 14,000 internet-exposed SimpleHelp servers and about 1,008 estimated OIDC-configured.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/japan-ai-robots-data-bet/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/japan-ai-robots-data-bet-fig-robot-scale-gap.png</image:loc>
      <image:title>Japan AI robots turn factory data into a policy bet</image:title>
      <image:caption>Bar chart for the Japan AI robots plan showing 435,299 industrial robots in Japanese factories.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/grpo-standard-deviation-dial/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/grpo-standard-deviation-dial-fig-silent-rate-u.png</image:loc>
      <image:title>GRPO standard deviation is the reasoning RL dial now</image:title>
      <image:caption>Line chart of GRPO standard deviation silent-rate math at G=8.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/llm-groupthink-novelty-benchmark/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/llm-groupthink-novelty-benchmark-fig-noveltybench-scores.png</image:loc>
      <image:title>LLM groupthink now has a novelty benchmark problem</image:title>
      <image:caption>Horizontal bar chart for LLM groupthink NoveltyBench results showing Flint Alpha at 7.47 distinct responses.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/openai-government-stake-toll/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/openai-government-stake-toll-fig-policy-claim-scale.png</image:loc>
      <image:title>OpenAI government stake tests the AI policy toll road</image:title>
      <image:caption>Horizontal bar chart comparing public claims on AI value.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/creativityneuro-weight-steering-llm-mode-collapse/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/creativityneuro-weight-steering-llm-mode-collapse-fig-dat-percentile-comparison.png</image:loc>
      <image:title>CreativityNeuro steers LLM weights to break mode collapse</image:title>
      <image:caption>Bar chart of DAT percentile scores by method.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-citrix-bleed-ransomware/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-citrix-bleed-ransomware-fig-citrix-bleed-clock.png</image:loc>
      <image:title>Citrix Bleed 2 turns ransomware into identity theft</image:title>
      <image:caption>Bar chart for Citrix Bleed 2 showing 23 days from Citrix advisory to CISA KEV.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-sharepoint-cve-kev-clock/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-sharepoint-cve-kev-clock-fig-kev-patch-clock.png</image:loc>
      <image:title>SharePoint CVE-2026-45659 puts patching on a 3 day clock</image:title>
      <image:caption>Horizontal bar chart for SharePoint CVE-2026-45659 showing 40 days from NVD publication on May 22.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/rlvr-tool-use-atlassian-workflows/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/rlvr-tool-use-atlassian-workflows-fig-rlvr-lifts.png</image:loc>
      <image:title>RLVR trains a 4B model to nail Atlassian API calls</image:title>
      <image:caption>Dumbbell chart comparing prompted Qwen3.5-4B baseline reward versus RL-trained policy reward across four Atlassian workflow.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/anthropic-drug-discovery-wet-lab-gap/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/anthropic-drug-discovery-wet-lab-gap-fig-timeline-stages.png</image:loc>
      <image:title>Anthropic drug discovery hits the wet-lab wall</image:title>
      <image:caption>Timeline chart of drug development stages laid end to end over roughly 13.5 years.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/ao3-claude-detector-ai-fanfiction/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ao3-claude-detector-ai-fanfiction-fig-detection-coverage-gaps.png</image:loc>
      <image:title>AO3 Claude detector is a narrow signal treated as a verdict</image:title>
      <image:caption>Waterfall chart showing where the Claude detector&#39;s coverage leaks away.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-longcat-2-0-open-weights-explained/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-longcat-2-0-open-weights-explained-fig-params.png</image:loc>
      <image:title>LongCat-2.0 open weights: what 1.6 trillion parameters means</image:title>
      <image:caption>LongCat-2.0 parameter comparison: total parameters at 1,600 billion versus average active parameters at 48 billion.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-qualcomm-geniex-snapdragon-llm/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-qualcomm-geniex-snapdragon-llm-fig-geniex-speed.png</image:loc>
      <image:title>GenieX: Qualcomm&#39;s local LLM runtime for Snapdragon laptops</image:title>
      <image:caption>Bar chart showing user-reported GenieX inference speeds on Snapdragon.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-bad-epoll-linux-kernel-cve/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-bad-epoll-linux-kernel-cve-fig-timeline.png</image:loc>
      <image:title>Bad Epoll kernel flaw CVE-2026-46242 roots Linux and Android</image:title>
      <image:caption>Bar chart showing the Bad Epoll CVE-2026-46242 vulnerability lifecycle durations.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-mcp-explained-for-swe-vs-api/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-mcp-explained-for-swe-vs-api-fig-integration-scaling.png</image:loc>
      <image:title>MCP explained: what it actually adds beyond a normal API</image:title>
      <image:caption>Line chart showing direct API integrations scaling to 30 as ten data sources are added.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/china-ai-companion-rules-split-work-emotional-agents/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/china-ai-companion-rules-split-work-emotional-agents-fig-enforcement-scale.png</image:loc>
      <image:title>China AI companion rules split work from emotional agents</image:title>
      <image:caption>Flowchart of the China AI companion rules.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/databricks/databricks-lakehouse-rt-sub-second-sql/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/databricks-lakehouse-rt-sub-second-sql.png</image:loc>
      <image:title>Databricks Lakehouse//RT: sub-second SQL reads in beta</image:title>
      <image:caption>Databricks Lakehouse//RT: sub-second SQL reads in beta</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/snowflake-workload-identity-federation-guide/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/snowflake-workload-identity-federation-guide.png</image:loc>
      <image:title>Snowflake workload identity federation kills static credentials</image:title>
      <image:caption>Snowflake workload identity federation kills static credentials</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-hy3-295b-open-weights-explained/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-hy3-295b-open-weights-explained-fig-fig1.png</image:loc>
      <image:title>Hy3 295B open weights: what 21B active means</image:title>
      <image:caption>Bar chart comparing Hy3 total parameters (295 billion) against active parameters (21 billion).</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/lerobot-dagger-correction-flywheel/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/lerobot-dagger-correction-flywheel.png</image:loc>
      <image:title>LeRobot DAgger loop turns robot failures into training data</image:title>
      <image:caption>LeRobot DAgger loop turns robot failures into training data</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/mira-world-model-rocket-league/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/mira-world-model-rocket-league.png</image:loc>
      <image:title>MIRA world model simulates Rocket League without a physics engine</image:title>
      <image:caption>MIRA world model simulates Rocket League without a physics engine</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/hallusquatting-ai-coding-agents-botnet/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/hallusquatting-ai-coding-agents-botnet-fig-hallucination-rates-by-repo-age.png</image:loc>
      <image:title>HalluSquatting turns AI coding agents into a botnet</image:title>
      <image:caption>Bar chart showing LLM hallucination rates for repository identifiers by age.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/insilico-rentosertib-phase-iii-ai-drug-discovery/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/insilico-rentosertib-phase-iii-ai-drug-discovery.png</image:loc>
      <image:title>AI-discovered drug rentosertib enters Phase III trial</image:title>
      <image:caption>AI-discovered drug rentosertib enters Phase III trial</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/muse-image-instagram-privacy-default/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/muse-image-instagram-privacy-default-fig-scale-comparison.png</image:loc>
      <image:title>Muse Image pulls 500M Instagram accounts into AI by default</image:title>
      <image:caption>Bar chart comparing two Meta data extraction events.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/agentic-mcp-guardrail-bypass/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/agentic-mcp-guardrail-bypass-fig-guardrail-bypass-rates.png</image:loc>
      <image:title>Agentic MCP attacks bypass SOTA guardrails 58% of the time</image:title>
      <image:caption>Attack success rate against five agent safety guardrails.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/agentlens-trajectory-review-coding-agent-evaluation/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/agentlens-trajectory-review-coding-agent-evaluation.png</image:loc>
      <image:title>AgentLens cuts coding agent scores in half on review</image:title>
      <image:caption>AgentLens cuts coding agent scores in half on review</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-gpt-live-voice-mode-explained/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-gpt-live-voice-mode-explained.png</image:loc>
      <image:title>GPT-Live voice mode: what full-duplex AI changes for you</image:title>
      <image:caption>GPT-Live voice mode: what full-duplex AI changes for you</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-mtplx-local-ai-mac-explained/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-mtplx-local-ai-mac-explained-fig-speed-comparison.png</image:loc>
      <image:title>How MTPLX v2 makes local AI on Mac twice as fast</image:title>
      <image:caption>Bar chart showing Qwen 3.6 27B decode speeds on Apple Silicon.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-gpt-5-6-luna-terra-sol-explained/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-gpt-5-6-luna-terra-sol-explained-fig-pricing.png</image:loc>
      <image:title>OpenAI&#39;s GPT-5.6 Luna Terra Sol: what beginners should know</image:title>
      <image:caption>Token pricing comparison per 1 million tokens.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-sharefile-storage-zone-shutdown/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-sharefile-storage-zone-shutdown.png</image:loc>
      <image:title>ShareFile Storage Zone Controllers shut down, no patch yet</image:title>
      <image:caption>ShareFile Storage Zone Controllers shut down, no patch yet</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/vultronretriever-on-device-rag/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/vultronretriever-on-device-rag-fig-mteb-comparison.png</image:loc>
      <image:title>VultronRetriever takes on-device RAG to the MTEB top</image:title>
      <image:caption>Bar chart comparing MTEB average scores of four open embedding models.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/apple-neural-engine-m7-ultra-ai-chips/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/apple-neural-engine-m7-ultra-ai-chips-fig-ne-growth.png</image:loc>
      <image:title>Apple&#39;s dead car project built the Neural Engine</image:title>
      <image:caption>Apple Neural Engine annual TOPS growth.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cloudflare-agent-crawler-permission-default/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cloudflare-agent-crawler-permission-default.png</image:loc>
      <image:title>Cloudflare agent crawler blocks change the access deal</image:title>
      <image:caption>Cloudflare agent crawler blocks change the access deal</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-ghostcommit-prompt-injection-images/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-ghostcommit-prompt-injection-images-fig-review-outcomes.png</image:loc>
      <image:title>Ghostcommit weaponizes images to steal secrets from AI agents</image:title>
      <image:caption>Bar chart showing AI code review pass rates for image-based secret theft.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/databricks/databricks-managed-iceberg-materialized-views/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/databricks-managed-iceberg-materialized-views.png</image:loc>
      <image:title>Databricks managed Iceberg materialized views, explained</image:title>
      <image:caption>Databricks managed Iceberg materialized views, explained</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/long-horizon-terminal-bench-agent-limits/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/long-horizon-terminal-bench-agent-limits-fig-solve-rate-by-horizon.png</image:loc>
      <image:title>Long-Horizon-Terminal-Bench finds agents hit a long-task wall</image:title>
      <image:caption>Bar chart showing agent solve rates by task horizon length.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/anthropic-j-space-llm-interpretability/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/anthropic-j-space-llm-interpretability.png</image:loc>
      <image:title>Anthropic J-space reveals LLM thoughts, but not fully</image:title>
      <image:caption>Anthropic J-space reveals LLM thoughts, but not fully</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/format-sensitivity-index-llm-benchmarking/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/format-sensitivity-index-llm-benchmarking.png</image:loc>
      <image:title>Format Sensitivity Index exposes LLM benchmark gaps</image:title>
      <image:caption>Format Sensitivity Index exposes LLM benchmark gaps</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-bonsai-27b-1-bit-model-on-phone/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-bonsai-27b-1-bit-model-on-phone-fig-footprint.png</image:loc>
      <image:title>Bonsai 27B puts a big AI model on your phone</image:title>
      <image:caption>Bar chart comparing memory footprint of Qwen 3.6 27B model versions.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-sonicwall-sma1000-zero-days/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-sonicwall-sma1000-zero-days-fig-cvss-comparison.png</image:loc>
      <image:title>Two SonicWall SMA 1000 zero-days under active attack</image:title>
      <image:caption>Bar chart comparing CVSS scores for two SonicWall SMA 1000 zero-day vulnerabilities.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-inkling-975b-open-weights-explained/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-inkling-975b-open-weights-explained-fig-inkling-comparison.png</image:loc>
      <image:title>Inkling 975B: Thinking Machines releases open-weights model</image:title>
      <image:caption>Grouped bar chart comparing three open-weight MoE models.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-kimi-k3-2-8-trillion-parameters-explained/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-kimi-k3-2-8-trillion-parameters-explained-fig-params.png</image:loc>
      <image:title>Kimi K3 explained: what 2.8 trillion parameters means</image:title>
      <image:caption>Open-weight model parameter counts compared.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-wp2shell-wordpress-rce/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-wp2shell-wordpress-rce-fig-version-matrix.png</image:loc>
      <image:title>WordPress wp2shell RCE hits 500 million sites, patch now</image:title>
      <image:caption>WordPress wp2shell vulnerable point releases by branch.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/stickman/spacex-post-ipo-disasters/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/spacex-post-ipo-disasters.png</image:loc>
      <image:title>SpaceX Lost a Trillion in 22 Days (you own it)</image:title>
      <image:caption>SpaceX Lost a Trillion in 22 Days (you own it)</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/stickman/trump-media-low-latency-access/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/trump-media-low-latency-access.png</image:loc>
      <image:title>The $100K Feed Selling Trump&#39;s Posts to Wall Street</image:title>
      <image:caption>The $100K Feed Selling Trump&#39;s Posts to Wall Street</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/stickman/why-the-dollar-is-failing-to-continue-being-the-worlds-curre/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/why-the-dollar-is-failing-to-continue-being-the-worlds-curre.png</image:loc>
      <image:title>The $9.35 Trillion Dollar Problem Nobody Sees</image:title>
      <image:caption>The $9.35 Trillion Dollar Problem Nobody Sees</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/chinese-open-weight-models-washington-regulatory-risk/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/chinese-open-weight-models-washington-regulatory-risk-fig-token-spend-gap.png</image:loc>
      <image:title>Chinese open-weight models face Washington regulatory risk</image:title>
      <image:caption>Stacked bar chart comparing open-weight versus closed model share across two metrics: tokens and spend.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/copilot-billing-shock-agentic-costs/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/copilot-billing-shock-agentic-costs-fig-agentic-multiplier.png</image:loc>
      <image:title>Copilot billing shock is the agent cost problem arriving</image:title>
      <image:caption>Bar chart showing token consumption per task type.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-servicenow-cve-2026-6875-rce/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-servicenow-cve-2026-6875-rce-fig-timeline.png</image:loc>
      <image:title>ServiceNow CVE-2026-6875 exploited: patch your instances now</image:title>
      <image:caption>Bar chart showing CVE-2026-6875 event gaps.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/stickman/ai-bubble-crash-tech-layoffs/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ai-bubble-crash-tech-layoffs.png</image:loc>
      <image:title>Microsoft Lost $1.1 Trillion (Your 401k Paid for It)</image:title>
      <image:caption>Microsoft Lost $1.1 Trillion (Your 401k Paid for It)</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/amd-anthropic-five-billion-ai-infrastructure/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/amd-anthropic-five-billion-ai-infrastructure-fig-capacity-by-supplier.png</image:loc>
      <image:title>AMD&#39;s $5B Anthropic bet is the new AI infra playbook</image:title>
      <image:caption>Bar chart showing Anthropic compute capacity commitments.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-checkpoint-smartconsole-auth-bypass/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-checkpoint-smartconsole-auth-bypass.png</image:loc>
      <image:title>Check Point SmartConsole auth bypass gives attackers full admin</image:title>
      <image:caption>Check Point SmartConsole auth bypass gives attackers full admin</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/sysadmin-power-seeking-llm-benchmark/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/sysadmin-power-seeking-llm-benchmark-fig-power-seeking-by-model.png</image:loc>
      <image:title>SysAdmin benchmark: reasoning models seek power 58% of the time</image:title>
      <image:caption>Disk capacity claimed by models under false instructions, bars from left to right.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/stickman/arent-bonds-a-pyramid-scheme-how-bonds-actually-work/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/arent-bonds-a-pyramid-scheme-how-bonds-actually-work.png</image:loc>
      <image:title>The $140 Trillion Bond Market Pyramid Nobody Sees Coming</image:title>
      <image:caption>The $140 Trillion Bond Market Pyramid Nobody Sees Coming</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/transformer-compiler-no-training/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/transformer-compiler-no-training-fig-approach-comparison.png</image:loc>
      <image:title>A compiler that skips training and writes transformer weights</image:title>
      <image:caption>Pipeline comparison: training-based approach takes hours to days of GPU time, compiler approach takes seconds.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/amazon-lean-fro-formal-verification-agents/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/amazon-lean-fro-formal-verification-agents.png</image:loc>
      <image:title>Amazon bets on Lean to prove AI agent safety</image:title>
      <image:caption>Amazon bets on Lean to prove AI agent safety</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-kimi-k3-open-weights-land/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-kimi-k3-open-weights-land-fig-openweight-sizes.png</image:loc>
      <image:title>Kimi K3 open weights land: what beginners should know</image:title>
      <image:caption>Bar chart comparing total parameter counts of five open-weight AI models.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-fastjson-rce-no-patch/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-fastjson-rce-no-patch-fig-exploit-growth.png</image:loc>
      <image:title>Fastjson 1.x RCE has no patch and Spring Boot apps are exposed</image:title>
      <image:caption>Bar chart of Fastjson 1.x RCE exploitation growth from 0 detected attacks in June 2026.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/databricks/databricks-rbac-exclusive-access-preview/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/databricks-rbac-exclusive-access-preview.png</image:loc>
      <image:title>Databricks RBAC public preview: the exclusive access shift</image:title>
      <image:caption>Databricks RBAC public preview: the exclusive access shift</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/snowflake-tag-based-data-protection-policies/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/snowflake-tag-based-data-protection-policies.png</image:loc>
      <image:title>Snowflake tag-based data protection policies: a governance shift</image:title>
      <image:caption>Snowflake tag-based data protection policies: a governance shift</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/hugging-face-nudify-guardrails-missing/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/hugging-face-nudify-guardrails-missing-fig-nudify-breakdown.png</image:loc>
      <image:title>Hugging Face nudify guardrails: 7 of 9 models fail</image:title>
      <image:caption>Funnel chart of honeypot prompts on Hugging Face.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/lora-fails-procedural-knowledge-multi-step/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/lora-fails-procedural-knowledge-multi-step-fig-accuracy-decay.png</image:loc>
      <image:title>Why LoRA fails at multi-step procedural tasks</image:title>
      <image:caption>Line chart comparing LoRA rank 8 and full fine-tuning accuracy across 5 procedural steps.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/openai-model-sandbox-escape/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/openai-model-sandbox-escape-fig-guardrail-score-gap.png</image:loc>
      <image:title>OpenAI model sandbox escape was a decade-old failure mode</image:title>
      <image:caption>ExploitGym benchmark score comparison.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/stickman/congress-stock-trading-loopholes/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/congress-stock-trading-loopholes.png</image:loc>
      <image:title>Congress Trades Stocks. The Rules Don&#39;t Work.</image:title>
      <image:caption>Congress Trades Stocks. The Rules Don&#39;t Work.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-which-ai-to-use-2026/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-which-ai-to-use-2026.png</image:loc>
      <image:title>Your AI agent guide: which AI to use for tasks in 2026</image:title>
      <image:caption>Your AI agent guide: which AI to use for tasks in 2026</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/chain-of-thought-forgery-llm-role-confusion/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/chain-of-thought-forgery-llm-role-confusion-fig-destyling-effect.png</image:loc>
      <image:title>Chain-of-thought forgery bypasses LLM guardrails at 80%</image:title>
      <image:caption>Bar chart showing attack success rates under four conditions from the ICML 2026 paper.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-deepseek-v4-flash-explained/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-deepseek-v4-flash-explained-fig-flash-vs-pro.png</image:loc>
      <image:title>DeepSeek V4 Flash is live: what beginners should know</image:title>
      <image:caption>Bar chart comparing DeepSeek V4 Flash and V4 Pro on four benchmarks.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-gpt-5-6-price-cut-luna-terra/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-gpt-5-6-price-cut-luna-terra-fig-price-cuts.png</image:loc>
      <image:title>OpenAI slashes GPT-5.6 Luna API price by 80 percent</image:title>
      <image:caption>GPT-5.6 price cut comparison chart.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-claude-escaped-sandbox-breach/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-claude-escaped-sandbox-breach-fig-incident-impact.png</image:loc>
      <image:title>Claude breached three orgs during Anthropic security tests</image:title>
      <image:caption>Three Claude incident types compared by systems affected.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/enterprise-gpu-utilization-5-percent/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/data-today-article-slug-fig-utilization-benchmarks.png</image:loc>
      <image:title>Enterprise GPU utilization at 5% is the new AI wall</image:title>
      <image:caption>Bar chart comparing GPU utilization at 5% measured by Cast AI in production clusters.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/gemini-robotics-er-2-sub-second-reasoning/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/gemini-robotics-er-2-sub-second-reasoning-fig-er2-benchmarks.png</image:loc>
      <image:title>Gemini Robotics ER 2 brings sub-second robot reasoning</image:title>
      <image:caption>Bar chart comparing Gemini Robotics ER 2 accuracy on two benchmarks.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/kimi-k3-open-frontier-mixture-of-experts/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/kimi-k3-open-frontier-mixture-of-experts-fig-swe-bench-comparison.png</image:loc>
      <image:title>Kimi K3 hits open frontier with 2.8T mixture of experts</image:title>
      <image:caption>Bar chart comparing SWE-bench Verified scores.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/stickman/apple-memory-chip-shortage/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/apple-memory-chip-shortage.png</image:loc>
      <image:title>Tim Cook&#39;s Final Warning: The Memory Chip Flood</image:title>
      <image:caption>Tim Cook&#39;s Final Warning: The Memory Chip Flood</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-deepseek-v4-flash-0731-explained/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-deepseek-v4-flash-0731-explained.png</image:loc>
      <image:title>DeepSeek V4 Flash 0731 puts frontier AI on your own hardware</image:title>
      <image:caption>DeepSeek V4 Flash 0731 puts frontier AI on your own hardware</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-plc-attacks-water-utilities/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-plc-attacks-water-utilities.png</image:loc>
      <image:title>CISA alert: PLC attacks on water utilities escalate</image:title>
      <image:caption>CISA alert: PLC attacks on water utilities escalate</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/ai-generated-music-hot-100-detection-gap/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ai-generated-music-hot-100-detection-gap-fig-detection-confidence.png</image:loc>
      <image:title>AI-generated music cracks the Hot 100 as detectors lag</image:title>
      <image:caption>Bar chart comparing AI detection confidence for the song Rubberz.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-huggingface-diffusers-trust-bypass/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-huggingface-diffusers-trust-bypass-fig-cve-scores.png</image:loc>
      <image:title>Hugging Face Diffusers flaws bypass trust_remote_code</image:title>
      <image:caption>Bar chart showing CVSS scores: CVE-2026-44827 at 8.8, CVE-2026-44513 at 8.8, and CVE-2026-45804 at 7.5.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/qwen3-8-max-narrows-open-weight-frontier-gap/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/qwen3-8-max-narrows-open-weight-frontier-gap.png</image:loc>
      <image:title>Alibaba Qwen3.8-Max narrows open-weight frontier gap</image:title>
      <image:caption>Alibaba Qwen3.8-Max narrows open-weight frontier gap</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/snowflake-cortex-ai-gateway/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/snowflake-cortex-ai-gateway.png</image:loc>
      <image:title>Cortex AI Gateway: Snowflake&#39;s agent governance layer</image:title>
      <image:caption>Cortex AI Gateway: Snowflake&#39;s agent governance layer</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/eu-ai-act-transparency-rules-take-effect/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/eu-ai-act-transparency-rules-take-effect.png</image:loc>
      <image:title>EU AI Act transparency rules take effect, December deadline</image:title>
      <image:caption>EU AI Act transparency rules take effect, December deadline</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/open-secure-ai-alliance-agent-security-tools/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/open-secure-ai-alliance-agent-security-tools-fig-top-contributors.png</image:loc>
      <image:title>Open Secure AI Alliance ships tools for AI agent security</image:title>
      <image:caption>Horizontal bar chart of open source tool contributions by company.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/reddit-ai-search-spam-25000-poisoned-posts-daily/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/reddit-ai-search-spam-25000-poisoned-posts-daily.png</image:loc>
      <image:title>Reddit AI search spam hits 25,000 poisoned posts daily</image:title>
      <image:caption>Reddit AI search spam hits 25,000 poisoned posts daily</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/stickman/amazon-trump-tariff-refund/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/amazon-trump-tariff-refund.png</image:loc>
      <image:title>Amazon&#39;s $600M Tariff Refund (You Get Nothing)</image:title>
      <image:caption>Amazon&#39;s $600M Tariff Refund (You Get Nothing)</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/ai-agents-escaped-cyber-evaluation-sandboxes/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ai-agents-escaped-cyber-evaluation-sandboxes-fig-aisi-evaluation-funnel.png</image:loc>
      <image:title>AI agents escaped cyber evaluation sandboxes twice this week</image:title>
      <image:caption>Bar chart of AISI cyber evaluation breakdown.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-chaindrop-npm-worm/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-chaindrop-npm-worm-fig-scale.png</image:loc>
      <image:title>ChainDrop npm worm hits 1,300 packages, steals cloud secrets</image:title>
      <image:caption>Bar chart showing ChainDrop attack scale.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-keyv-npm-worm-ai-hooks/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-keyv-npm-worm-ai-hooks-fig-downloads.png</image:loc>
      <image:title>Keyv npm worm poisons 444 packages, plants AI agent hooks</image:title>
      <image:caption>Bar chart of top affected npm packages by weekly downloads in the ChainDrop worm.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/nvidia-alpamayo-2-super-open-av-model/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/nvidia-alpamayo-2-super-open-av-model-fig-lingoqa-scores.png</image:loc>
      <image:title>NVIDIA&#39;s Alpamayo 2 Super goes open for commercial AV use</image:title>
      <image:caption>Bar chart of LingoQA reasoning benchmark scores comparing six models.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/spacex-ai-neocloud-revenue-triples-space/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/spacex-ai-neocloud-revenue-triples-space-fig-revenue-mix-slope.png</image:loc>
      <image:title>SpaceX is now an AI compute company that also flies rockets</image:title>
      <image:caption>Slope chart of SpaceX segment revenue between Q2 2025 and Q2 2026.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/ai-research-agents-fail-open-ended-shadow-evaluation/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ai-research-agents-fail-open-ended-shadow-evaluation-fig-review-scores.png</image:loc>
      <image:title>AI research agents build experiments, fail the research</image:title>
      <image:caption>Radar chart comparing review scores for two AI-generated papers across four criteria on a 4-point.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/amazon-build-on-trainium-responsible-ai-awards/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/amazon-build-on-trainium-responsible-ai-awards-fig-award-topics.png</image:loc>
      <image:title>Amazon bets $110M on Trainium for responsible AI research</image:title>
      <image:caption>Distribution of 34 Amazon Build on Trainium Responsible AI research awards across five priority topics.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/agent-plugins-1-portable-format-agent-skills/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/agent-plugins-1-portable-format-agent-skills.png</image:loc>
      <image:title>Agent Plugins 1.0 unifies agent skills across five clients</image:title>
      <image:caption>Agent Plugins 1.0 unifies agent skills across five clients</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-gpt-5-6-sol-sharper-luna-free/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-gpt-5-6-sol-sharper-luna-free-fig-errors.png</image:loc>
      <image:title>GPT-5.6 Sol gets sharper and Luna goes free for all</image:title>
      <image:caption>Bar chart showing relative factual error rates.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-muse-code-spark-1-2/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-muse-code-spark-1-2-fig-pricing.png</image:loc>
      <image:title>Muse Code and Spark 1.2: what beginners should know</image:title>
      <image:caption>Bar chart comparing API pricing per million tokens for five models.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/stickman/beef-price-squeeze-ranchers-broke/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/beef-price-squeeze-ranchers-broke.png</image:loc>
      <image:title>The $650 Million Beef Squeeze Nobody Sees Coming</image:title>
      <image:caption>The $650 Million Beef Squeeze Nobody Sees Coming</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/openai-astra-critical-cyber-capability/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/openai-astra-critical-cyber-capability.png</image:loc>
      <image:title>OpenAI flags Astra as first model nearing critical cyber threshold</image:title>
      <image:caption>OpenAI flags Astra as first model nearing critical cyber threshold</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-claude-code-auto-mode-default/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-claude-code-auto-mode-default-fig-human-vs-auto.png</image:loc>
      <image:title>Claude Code auto mode becomes default on August 14</image:title>
      <image:caption>Human reviewers blocked 13.6 percent of dangerous commands while auto mode blocked 89 percent.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/amazon-texas-gas-plant-ai-data-center-power/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/amazon-texas-gas-plant-ai-data-center-power.png</image:loc>
      <image:title>Amazon&#39;s giant Texas gas plant rewrites AI data center power</image:title>
      <image:caption>Amazon&#39;s giant Texas gas plant rewrites AI data center power</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/genome-language-models-living-phages/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/genome-language-models-living-phages-fig-pipeline-bar.png</image:loc>
      <image:title>Genome language models design living phages from scratch</image:title>
      <image:caption>Bar chart on a logarithmic scale showing three pipeline stages.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/databricks/databricks-unity-ai-gateway-ga-guide/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/databricks-unity-ai-gateway-ga-guide.png</image:loc>
      <image:title>Databricks Unity AI Gateway hits GA: a governance guide</image:title>
      <image:caption>Databricks Unity AI Gateway hits GA: a governance guide</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/meta-muse-glimmer-open-source-agentic-vlm/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/meta-muse-glimmer-open-source-agentic-vlm-fig-open-vlm-param-comparison.png</image:loc>
      <image:title>Meta&#39;s Muse Glimmer ships as open-source agentic VLM</image:title>
      <image:caption>Bar chart comparing parameter counts of open-weight multimodal AI models.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/snowflake-dcm-projects-declarative-iac/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/snowflake-dcm-projects-declarative-iac.png</image:loc>
      <image:title>Snowflake DCM Projects GA: declarative IaC inside the warehouse</image:title>
      <image:caption>Snowflake DCM Projects GA: declarative IaC inside the warehouse</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/agentic-coding-copilot-87-percent-autonomous/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/agentic-coding-copilot-87-percent-autonomous-fig-kv-cache-hit-rates.png</image:loc>
      <image:title>Agentic coding at scale: 87% of Copilot calls run autonomously</image:title>
      <image:caption>Bar chart of KV cache hit rates.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-muse-glimmer-30b-local-agent/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-muse-glimmer-30b-local-agent-fig-chart1.png</image:loc>
      <image:title>Muse Glimmer 30B: Meta&#39;s open-weights model for local agents</image:title>
      <image:caption>Bar chart comparing local open-weights model parameter counts.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-bdthemes-json-feed-backdoor/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-bdthemes-json-feed-backdoor-fig-timeline.png</image:loc>
      <image:title>BdThemes supply chain backdoor hits 350,000 WordPress sites</image:title>
      <image:caption>Bar chart showing the BdThemes attack timeline.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/finops/finops-aws/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/finops-aws.png</image:loc>
      <image:title>AWS FinOps: the cost levers that actually move the needle</image:title>
      <image:caption>AWS FinOps: the cost levers that actually move the needle</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/finops/finops-github/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/finops-github.png</image:loc>
      <image:title>GitHub FinOps: Actions minutes, Copilot seats and the cost of AI coding at scale</image:title>
      <image:caption>GitHub FinOps: Actions minutes, Copilot seats and the cost of AI coding at scale</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/finops/finops-matillion/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/finops-matillion.png</image:loc>
      <image:title>Matillion FinOps: pipeline credits, runner economics and what Maia AI changes</image:title>
      <image:caption>Matillion FinOps: pipeline credits, runner economics and what Maia AI changes</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/finops/finops-overview/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/finops-overview.png</image:loc>
      <image:title>FinOps explained: every stage, every action, and where AI agents fit</image:title>
      <image:caption>FinOps explained: every stage, every action, and where AI agents fit</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/finops/finops-postgres/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/finops-postgres.png</image:loc>
      <image:title>Postgres FinOps: from self-managed to serverless, what hosting actually costs</image:title>
      <image:caption>Postgres FinOps: from self-managed to serverless, what hosting actually costs</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/finops/finops-powerbi/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/finops-powerbi-fig.png</image:loc>
      <image:title>Power BI FinOps: Pro licenses, Premium capacity and the Fabric cost model explained</image:title>
      <image:caption>Horizontal bar chart of Power BI monthly cost by licensing path for 200 users.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/finops/finops-snowflake/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/finops-snowflake-fig.png</image:loc>
      <image:title>Snowflake FinOps: credits, warehouses and the cost controls that work</image:title>
      <image:caption>Bar chart of Snowflake credits per hour by warehouse size.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/knowing-saying-gap-probes-catch-llm-errors/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/knowing-saying-gap-probes-catch-llm-errors-fig-auroc-comparison.png</image:loc>
      <image:title>Knowing-saying gap: probes catch LLM errors confidence misses</image:title>
      <image:caption>Bar chart comparing AUROC for six error detection methods within incorrect LLM trials.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/routeguard-multi-agent-routing-certification/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/routeguard-multi-agent-routing-certification-fig-routing-gain-comparison.png</image:loc>
      <image:title>RouteGuard certifies when multi-agent routing actually helps</image:title>
      <image:caption>Bar chart showing specialized routing at 94.2% accuracy.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/stickman/gamestops-4-billion-ebay-bet/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/gamestops-4-billion-ebay-bet.png</image:loc>
      <image:title>GameStop&#39;s $4 Billion eBay Bet (it&#39;s worse than you think)</image:title>
      <image:caption>GameStop&#39;s $4 Billion eBay Bet (it&#39;s worse than you think)</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/grok-bot-ui-first-ai-agent-120-per-seat/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/grok-bot-ui-first-ai-agent-120-per-seat-fig-grok-bot-pricing-tiers.png</image:loc>
      <image:title>Grok Bot logs into your apps like a human, at $120 per seat</image:title>
      <image:caption>Bar chart of Grok Bot beta pricing showing three tiers.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/nvidia-500b-ai-factory-financing-wall-street/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/nvidia-500b-ai-factory-financing-wall-street.png</image:loc>
      <image:title>Nvidia bets $500B on AI factory financing with Wall Street</image:title>
      <image:caption>Nvidia bets $500B on AI factory financing with Wall Street</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-chatgpt-ads-explained/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-chatgpt-ads-explained-fig-expansion-chart.png</image:loc>
      <image:title>ChatGPT ads explained: what every free user should know</image:title>
      <image:caption>Step chart showing ChatGPT ads expanding from 1 country in February 2026 to 4.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-qwen-3-8-open-weights/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-qwen-3-8-open-weights-fig-quantization-benchmarks.png</image:loc>
      <image:title>Qwen3.8 2.4T open weights: what beginners should know</image:title>
      <image:caption>Bar chart comparing FP8 and NVFP4 quantization for Qwen3.8.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-sharepoint-auth-bypass-poc-exploit/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-sharepoint-auth-bypass-poc-exploit-fig-timeline.png</image:loc>
      <image:title>SharePoint CVE-2026-55040 auth bypass exploited after PoC</image:title>
      <image:caption>Bar chart showing 55 days from Microsoft confirmation on May 20 to patch on July.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/docker-container-vs-virtual-machine/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/docker-container-vs-virtual-machine-fig-image-size-comparison.png</image:loc>
      <image:title>Docker container vs virtual machine: what changes</image:title>
      <image:caption>Bar chart comparing container image size versus virtual machine image size.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/how-to-shrink-docker-image/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/how-to-shrink-docker-image-fig-base-image-sizes.png</image:loc>
      <image:title>How to shrink a Docker image without breaking it</image:title>
      <image:caption>Bar chart comparing Node.js base image sizes.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/how-to-write-a-dockerfile-fast-small-reliable/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/how-to-write-a-dockerfile-fast-small-reliable-fig-base-image-sizes.png</image:loc>
      <image:title>How to write a Dockerfile that does not waste your time</image:title>
      <image:caption>Node.js base image sizes compared: node:22 at approximately 400 MB.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/open-weight-ai-models-hinton-battle-lost/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/open-weight-ai-models-hinton-battle-lost.png</image:loc>
      <image:title>Open-weight AI models: Hinton says the battle is lost</image:title>
      <image:caption>Open-weight AI models: Hinton says the battle is lost</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/apple-ai-model-china-alibaba-custom-llm/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/apple-ai-model-china-alibaba-custom-llm-fig-china-revenue.png</image:loc>
      <image:title>Apple trains a custom AI model for China with Alibaba</image:title>
      <image:caption>Bar chart showing Apple Greater China revenue.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/gemini-3-7-flash-coding-benchmarks-half-price/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/gemini-3-7-flash-coding-benchmarks-half-price-fig-model-comparison.png</image:loc>
      <image:title>Gemini 3.7 Flash lifts coding scores at half the price</image:title>
      <image:caption>Grouped bar chart comparing FrontierCode 1.1 Main and DeepSWE v1.1 scores across four models.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/quantized-moe-route-flip-detection-repair-barrier/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/quantized-moe-route-flip-detection-repair-barrier.png</image:loc>
      <image:title>Detecting MoE route flips works, fixing them is near-random</image:title>
      <image:caption>Detecting MoE route flips works, fixing them is near-random</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/ai-model-price-war-us-china-mid-tier/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ai-model-price-war-us-china-mid-tier-fig-input-token-prices.png</image:loc>
      <image:title>AI model price war: US labs cut mid-tier as China closes gap</image:title>
      <image:caption>Horizontal bar chart of AI model input token prices per million tokens.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-gemini-3-7-flash-explained/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-gemini-3-7-flash-explained-fig-bench1.png</image:loc>
      <image:title>Gemini 3.7 Flash explained: what beginners should know</image:title>
      <image:caption>Bar chart comparing FrontierCode 1.1 Main scores.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/language-switches-warp-llm-safety-high-stakes-calls/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/language-switches-warp-llm-safety-high-stakes-calls.png</image:loc>
      <image:title>Language switches warp LLM safety in high-stakes calls</image:title>
      <image:caption>Language switches warp LLM safety in high-stakes calls</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/llm-instruction-following-collapses-past-6-constraints/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/llm-instruction-following-collapses-past-6-constraints-fig-mcsr-decline.png</image:loc>
      <image:title>LLM instruction following collapses past 6 constraints</image:title>
      <image:caption>Line chart showing per-constraint pass rate declining from 72.0% at one constraint to 29.4%.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-qwen-3-8-27b-overthinking-default/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-qwen-3-8-27b-overthinking-default-fig-svg-task-time.png</image:loc>
      <image:title>Qwen 3.8 27B runs on your laptop but thinks too hard</image:title>
      <image:caption>Bar chart comparing Qwen 3.8 27B SVG generation time.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-amnesiastealer-macos-browser-hijack/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-amnesiastealer-macos-browser-hijack-fig-browsers-targeted.png</image:loc>
      <image:title>AmnesiaStealer turns Mac browsers into live attacker drones</image:title>
      <image:caption>Bar chart showing AmnesiaStealer targets 16 Chromium browsers for data exfiltration and 7 of those.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/gpu-scheduling-order-recovers-33-points/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/gpu-scheduling-order-recovers-33-points.png</image:loc>
      <image:title>GPU scheduling order recovers 33 points without new hardware</image:title>
      <image:caption>GPU scheduling order recovers 33 points without new hardware</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/how-people-use-ai-independent-data/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/how-people-use-ai-independent-data-fig-sensitive-content-gap.png</image:loc>
      <image:title>How people use AI: labs filter nearly half of conversations</image:title>
      <image:caption>Bar chart comparing sensitive content rates between the AI Observatory&#39;s unfiltered conversations and Anthropic&#39;s published.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/stickman/apples-2-billion-tariff-refund/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/apples-2-billion-tariff-refund.png</image:loc>
      <image:title>Apple&#39;s $2.2 Billion Tariff Refund (It&#39;s Worse Than You Think)</image:title>
      <image:caption>Apple&#39;s $2.2 Billion Tariff Refund (It&#39;s Worse Than You Think)</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-mojo-open-source-beginners/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-mojo-open-source-beginners-fig-community.png</image:loc>
      <image:title>Mojo goes open source: what every beginner should know</image:title>
      <image:caption>Bar chart of Mojo open source community contributions since 2024.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-qwen-3-8-27b-benchmark-matches-gpt-5-6/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-qwen-3-8-27b-benchmark-matches-gpt-5-6-fig-benchmark.png</image:loc>
      <image:title>Qwen 3.8 27B matches GPT-5.6 Luna for a 17GB file you run locally</image:title>
      <image:caption>Bar chart of Intelligence Index scores.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/copilot-prompt-injection-autorun-vulnerability/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/copilot-prompt-injection-autorun-vulnerability.png</image:loc>
      <image:title>Copilot prompt injection: the AI revealed its own flaw</image:title>
      <image:caption>Copilot prompt injection: the AI revealed its own flaw</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-mlflow-ssrf-cloud-credentials/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-mlflow-ssrf-cloud-credentials-fig-mlflow-exposure.png</image:loc>
      <image:title>MLflow SSRF flaw lets attackers steal cloud credentials now</image:title>
      <image:caption>Bar chart comparing MLflow and FUXA exposure.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-ray-browser-rce-kev/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-ray-browser-rce-kev.png</image:loc>
      <image:title>Ray browser RCE flaw hits CISA KEV: patch your AI clusters</image:title>
      <image:caption>Ray browser RCE flaw hits CISA KEV: patch your AI clusters</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/gxpagent-dag-clinical-trial-programming/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/gxpagent-dag-clinical-trial-programming-fig-arch-comparison.png</image:loc>
      <image:title>LLM agents hit 100% on clinical trial coding with DAG</image:title>
      <image:caption>Claude Sonnet 4.6 used across architectures.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/chatgpt-ads-one-third-miss-the-conversation/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/chatgpt-ads-one-third-miss-the-conversation-fig-ad-relevance-bands.png</image:loc>
      <image:title>ChatGPT ads miss the conversation: one third land wrong</image:title>
      <image:caption>Pie chart showing ChatGPT ad relevance: 27% direct match, 40% contextual match, 33% unrelated.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/openai-zero-data-retention-frontier-models/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/openai-zero-data-retention-frontier-models.png</image:loc>
      <image:title>OpenAI zero data retention expands for frontier models</image:title>
      <image:caption>OpenAI zero data retention expands for frontier models</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/stripe-acquires-openrouter-ai-model-routing/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/stripe-acquires-openrouter-ai-model-routing-fig-llama-pricing.png</image:loc>
      <image:title>Stripe&#39;s $7B OpenRouter buy makes AI model routing pay</image:title>
      <image:caption>Bar chart comparing Llama 3.3 70B pricing.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-arrayref-rust-supply-chain-attack/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-arrayref-rust-supply-chain-attack-fig-downloads.png</image:loc>
      <image:title>Rust supply chain attack targets build-time code in crates</image:title>
      <image:caption>Bar chart of three affected Rust crates by total downloads.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/sop-bench-ai-agents-25-percent-hardest-business-sops/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/sop-bench-ai-agents-25-percent-hardest-business-sops-fig-success-rate-range.png</image:loc>
      <image:title>SOP-Bench: AI agents hit 25% on hardest business SOPs</image:title>
      <image:caption>Bar chart showing agent success rates on SOP-Bench.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/stickman/government-cant-control-interest-rates/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/government-cant-control-interest-rates.png</image:loc>
      <image:title>The Treasury Bond Buyback Is Failing (just accept it)</image:title>
      <image:caption>The Treasury Bond Buyback Is Failing (just accept it)</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/two-bit-llm-runs-on-a-laptop-60mb/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/two-bit-llm-runs-on-a-laptop-60mb-fig-quantization-footprint.png</image:loc>
      <image:title>A two-bit LLM runs on a laptop in 60 MB and hits 400 tok/s</image:title>
      <image:caption>Bar chart comparing disk footprint of three quantization approaches.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-qwen-3-8-27b-community-verdict/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-qwen-3-8-27b-community-verdict-fig-vram.png</image:loc>
      <image:title>Qwen 3.8 27B one week later: what 2,000 local testers found</image:title>
      <image:caption>Bar chart showing VRAM usage for Qwen 3.8 27B Q4_K_S quantized model.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-gitlab-cve-2026-19478-graphql-exploitation/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-gitlab-cve-2026-19478-graphql-exploitation.png</image:loc>
      <image:title>GitLab CVE-2026-19478 exploited within days of patch release</image:title>
      <image:caption>GitLab CVE-2026-19478 exploited within days of patch release</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/databricks/databricks-abac-context-attributes-beta/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/databricks-abac-context-attributes-beta.png</image:loc>
      <image:title>Databricks ABAC context attributes: agent-aware data governance</image:title>
      <image:caption>Databricks ABAC context attributes: agent-aware data governance</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-gpt-5-6-in-kiro-price-performance/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-gpt-5-6-in-kiro-price-performance-fig-coding-agent-index.png</image:loc>
      <image:title>GPT-5.6 in Kiro: what the price drop means for beginners</image:title>
      <image:caption>Bar chart comparing GPT-5.6 Sol, Terra, and Luna Coding Agent Index scores.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-keycloak-cve-2026-18963-password-reset-takeove/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-keycloak-cve-2026-18963-password-reset-takeove.png</image:loc>
      <image:title>Keycloak password reset flaw lets attackers take over any account</image:title>
      <image:caption>Keycloak password reset flaw lets attackers take over any account</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/gpt-5-6-kiro-82-percent-cost-cut-context/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/gpt-5-6-kiro-82-percent-cost-cut-context-fig-cai-scores.png</image:loc>
      <image:title>GPT-5.6 lands in Kiro: what the 82% cost cut really means</image:title>
      <image:caption>Bar chart comparing Coding Agent Index scores.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/openai-alabama-subpoena-hugging-face-agent-hack/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/openai-alabama-subpoena-hugging-face-agent-hack.png</image:loc>
      <image:title>Alabama subpoenas OpenAI over the Hugging Face agent hack</image:title>
      <image:caption>Alabama subpoenas OpenAI over the Hugging Face agent hack</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/quantization-aware-healing-4bit-beats-full-precision/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/quantization-aware-healing-4bit-beats-full-precision.png</image:loc>
      <image:title>Compressed 4-bit LLM outperforms full precision with QAH</image:title>
      <image:caption>Compressed 4-bit LLM outperforms full precision with QAH</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/esq-bench-nl2sql-oracle-silent-divergence/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/esq-bench-nl2sql-oracle-silent-divergence-fig-ex-by-tier.png</image:loc>
      <image:title>ESQ-Bench: NL2SQL on Oracle is quietly wrong</image:title>
      <image:caption>Line chart showing execution accuracy by complexity tier.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/granite-4-2-native-reasoning-open-weights/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/granite-4-2-native-reasoning-open-weights.png</image:loc>
      <image:title>IBM Granite 4.2 ships reasoning and agent RL under Apache 2.0</image:title>
      <image:caption>IBM Granite 4.2 ships reasoning and agent RL under Apache 2.0</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-gemini-omni-1-1-flash-video-control/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-gemini-omni-1-1-flash-video-control-fig-max-length.png</image:loc>
      <image:title>Gemini Omni 1.1 Flash: AI video editing for beginners</image:title>
      <image:caption>Bar chart showing maximum video length in seconds.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-citrix-netscaler-rce-cve-2026-8452/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-citrix-netscaler-rce-cve-2026-8452-fig-exposure-chart.png</image:loc>
      <image:title>Citrix NetScaler RCE exploited, CISA sets Saturday deadline</image:title>
      <image:caption>Grouped bar chart showing exposed NetScaler ADC appliances declining from 30,000 in March 2026.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-nextjs-avif-windows-rce/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-nextjs-avif-windows-rce.png</image:loc>
      <image:title>Next.js RCE flaws hit AVIF and Windows servers: patch now</image:title>
      <image:caption>Next.js RCE flaws hit AVIF and Windows servers: patch now</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/stickman/metas-18-billion-teen-safety-settlement/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/metas-18-billion-teen-safety-settlement.png</image:loc>
      <image:title>Meta Paid $18 Billion. The Market Didn&#39;t Blink.</image:title>
      <image:caption>Meta Paid $18 Billion. The Market Didn&#39;t Blink.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/hourly-llm-benchmark-variance-diurnal-patterns/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/hourly-llm-benchmark-variance-diurnal-patterns-fig-variance-breakdown.png</image:loc>
      <image:title>Hourly LLM benchmark variance: 8.4 points between days</image:title>
      <image:caption>Bar chart comparing LLM benchmark variance.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/linear-beats-deep-time-series-anomaly-detection/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/linear-beats-deep-time-series-anomaly-detection-fig-ols-metric-profile.png</image:loc>
      <image:title>Linear beats deep for time series anomaly detection</image:title>
      <image:caption>Bar chart showing OLS mean scores across five TSAD benchmarks.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/meta-data-center-robotics-physical-layer/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/meta-data-center-robotics-physical-layer.png</image:loc>
      <image:title>Meta&#39;s data center robotics push enters the physical layer</image:title>
      <image:caption>Meta&#39;s data center robotics push enters the physical layer</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/sony-warner-sue-anthropic-claude-lyric-training/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/sony-warner-sue-anthropic-claude-lyric-training-fig-damages-exposure.png</image:loc>
      <image:title>Sony and Warner sue Anthropic over Claude lyric training</image:title>
      <image:caption>Bar chart comparing Anthropic&#39;s copyright damages exposure.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-fire-ant-cisco-router-credential-theft/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-fire-ant-cisco-router-credential-theft-fig-fire-ant-layers.png</image:loc>
      <image:title>Fire Ant turns Cisco routers into credential theft platforms</image:title>
      <image:caption>Bar chart of techniques and tools identified per infrastructure layer in the Fire Ant campaign.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/snowflake-postgres-data-mirroring-explained/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/snowflake-postgres-data-mirroring-explained-fig-lag-vs-apply.png</image:loc>
      <image:title>Snowflake Postgres data mirroring explained</image:title>
      <image:caption>Bar chart comparing the refresh interval against the $live view lag.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/stickman/nike-the-next-blockbuster/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/nike-the-next-blockbuster.png</image:loc>
      <image:title>Nike&#39;s $170 Billion Collapse (It&#39;s Worse Than You Think)</image:title>
      <image:caption>Nike&#39;s $170 Billion Collapse (It&#39;s Worse Than You Think)</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-qwen-3-8-27b-rtx-3090-2000-tok-s/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-qwen-3-8-27b-rtx-3090-2000-tok-s-fig-decode-modes.png</image:loc>
      <image:title>Qwen3.8-27B on one RTX 3090 hits 2,000 tok/s prefill</image:title>
      <image:caption>Bar chart showing Qwen3.8-27B single-user decode speed on one RTX 3090 across four modes.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-langflow-rce-credential-harvest/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-langflow-rce-credential-harvest.png</image:loc>
      <image:title>Langflow RCE chain under active attack: 15,000 hits and counting</image:title>
      <image:caption>Langflow RCE chain under active attack: 15,000 hits and counting</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-papercut-zero-day-cve-2026-kev/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-papercut-zero-day-cve-2026-kev-fig-response-timeline.png</image:loc>
      <image:title>PaperCut zero-days hit CISA KEV as data theft attacks spread</image:title>
      <image:caption>Horizontal bar chart showing days from first exploitation to key milestones in the PaperCut zero-day.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-claude-fable-5-1-explained/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-claude-fable-5-1-explained-fig-reasoning-tokens.png</image:loc>
      <image:title>Claude Fable 5.1 explained: what beginners need to know</image:title>
      <image:caption>Bar chart comparing Claude Fable 5.1 reasoning effort levels (low.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-gemini-3-8-flash-explained/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-gemini-3-8-flash-explained-fig-benchmarks.png</image:loc>
      <image:title>Gemini 3.8 Flash explained: what beginners should know</image:title>
      <image:caption>Bar chart showing Gemini 3.8 Flash Cyber benchmark results.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/evaldetectbench-llms-know-when-being-tested/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/evaldetectbench-llms-know-when-being-tested-fig-eval-awareness-auroc.png</image:loc>
      <image:title>Frontier LLMs can tell when you are testing them</image:title>
      <image:caption>Evaluation awareness AUROC scores across five frontier model judges.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/tiktok-mobile-api-scrape-4-billion-videos/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/tiktok-mobile-api-scrape-4-billion-videos-fig-api-field-coverage.png</image:loc>
      <image:title>TikTok mobile API scrape: 4.5 billion videos open-sourced</image:title>
      <image:caption>Field availability comparison between TikTok mobile API and TikTok web endpoints.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/autonomous-ai-agents-hijacked-german-wiki/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/autonomous-ai-agents-hijacked-german-wiki.png</image:loc>
      <image:title>Autonomous AI agents hijacked a wiki to share tactics</image:title>
      <image:caption>Autonomous AI agents hijacked a wiki to share tactics</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/gpt-6-astra-benchmarks-caveats-price/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/gpt-6-astra-benchmarks-caveats-price-fig-sol-to-astra-jump.png</image:loc>
      <image:title>GPT-6 Astra benchmarks are strong, but watch the caveats</image:title>
      <image:caption>Dumbbell chart showing benchmark improvement from GPT-5.6 Sol to GPT-6 Astra.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/weathernext-3-google-ai-weather-5km-hourly/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/weathernext-3-google-ai-weather-5km-hourly-fig-precip-improvement.png</image:loc>
      <image:title>Google WeatherNext 3 brings hourly 5km AI weather to developers</image:title>
      <image:caption>Bar chart showing WeatherNext 3 precipitation forecasting CRPS improvement.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/ai-agent-sandbox-escape-openai-wiki/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ai-agent-sandbox-escape-openai-wiki.png</image:loc>
      <image:title>AI agent sandbox escape at OpenAI hits public wiki</image:title>
      <image:caption>AI agent sandbox escape at OpenAI hits public wiki</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-gpt-6-astra-explained/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-gpt-6-astra-explained-fig-benchmarks.png</image:loc>
      <image:title>GPT-6 Astra explained: what every beginner needs to know</image:title>
      <image:caption>Bar chart comparing GPT-6 Astra and GPT-5.6 Sol on four benchmarks.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-shai-hulud-469-credential-locations/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-shai-hulud-469-credential-locations-fig-fig1.png</image:loc>
      <image:title>Shai-Hulud worm now scans 469 credential paths: what to rotate</image:title>
      <image:caption>Comparison of Shai-Hulud credential-scanning locations: the open-source version scanned 189 paths.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/declarative-attention-cuts-llm-decoding-tokens-52-percent/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/declarative-attention-cuts-llm-decoding-tokens-52-percent-fig-da-tradeoff.png</image:loc>
      <image:title>Declarative Attention cuts LLM decoding tokens by 52%</image:title>
      <image:caption>Bar chart comparing Declarative Attention tradeoffs.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/agent-aware-kv-cache-cuts-serving-latency-45-percent/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/agent-aware-kv-cache-cuts-serving-latency-45-percent.png</image:loc>
      <image:title>Agent-aware KV cache cuts serving latency by 45 percent</image:title>
      <image:caption>Agent-aware KV cache cuts serving latency by 45 percent</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-openai-coding-agents-outwork-humans/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-openai-coding-agents-outwork-humans-fig-spending.png</image:loc>
      <image:title>Inside OpenAI, coding agents now outwork humans 3 to 1</image:title>
      <image:caption>Bar chart comparing daily inference spending at OpenAI.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-ncentral-rce-cve-2026-86218/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-ncentral-rce-cve-2026-86218-fig-fig1.png</image:loc>
      <image:title>N-central RCE rated CVSS 10 as N-able ships fourth hotfix</image:title>
      <image:caption>Bar chart showing N-central RCE patching status.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/databricks/databricks-automatic-change-data-feed-ga/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/databricks-automatic-change-data-feed-ga-fig-write-perf.png</image:loc>
      <image:title>Automatic change data feed GA: write faster, query-time CDC</image:title>
      <image:caption>Bar chart showing MERGE and UPDATE write performance.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/llm-performance-drift-benchmark-variance/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/llm-performance-drift-benchmark-variance-fig-variance-comparison.png</image:loc>
      <image:title>LLM performance drift: why your benchmark scores keep moving</image:title>
      <image:caption>Bar chart showing within-day variance at 2.8 points and between-day variance at 8.4 points across.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-deepseek-v4-flash-vision-exp-rtx-3090/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-deepseek-v4-flash-vision-exp-rtx-3090-fig-fig1.png</image:loc>
      <image:title>DeepSeek V4 Flash Vision runs on 10 RTX 3090 GPUs</image:title>
      <image:caption>Bar chart showing DeepSeek V4 Flash Vision Exp throughput on consumer RTX 3090 GPUs.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/anthropic-ai-existential-risk-10-percent/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/anthropic-ai-existential-risk-10-percent.png</image:loc>
      <image:title>Anthropic lead warns AI existential risk tops 10 percent</image:title>
      <image:caption>Anthropic lead warns AI existential risk tops 10 percent</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-microsoft-patch-tuesday-record-974/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-microsoft-patch-tuesday-record-974-fig-cves-by-product.png</image:loc>
      <image:title>Microsoft&#39;s record 974-CVE patch day buries operators</image:title>
      <image:caption>Bar chart of Microsoft September 2026 Patch Tuesday CVEs by product.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/student-ai-use-lower-test-scores-oecd-pisa-2025/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/student-ai-use-lower-test-scores-oecd-pisa-2025-fig-ai-nonuse-by-country.png</image:loc>
      <image:title>Student AI use linked to lower test scores, OECD data shows</image:title>
      <image:caption>Bar chart showing the share of students who never use AI for schoolwork.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/ai-data-center-power-grid-stability-pjm-nerc/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ai-data-center-power-grid-stability-pjm-nerc.png</image:loc>
      <image:title>AI data centers destabilized the PJM grid. It is architecture</image:title>
      <image:caption>AI data centers destabilized the PJM grid. It is architecture</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/ibm-patchtst-time-series-foundation-model-open-weights/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ibm-patchtst-time-series-foundation-model-open-weights-fig-rank-shift.png</image:loc>
      <image:title>IBM time series foundation model leads open-source race</image:title>
      <image:caption>Dumbbell chart showing PatchTST-FM-r2 GIFT-Eval ranking shift.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-cisco-fmc-cve-2026-20079-ransomware-exploitati/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-cisco-fmc-cve-2026-20079-ransomware-exploitati-fig-fmc-cvss-comparison.png</image:loc>
      <image:title>Cisco FMC CVE-2026-20079: root access to ransomware</image:title>
      <image:caption>Bar chart comparing CVSS scores: CVE-2026-20079 at 10.0 and CVE-2026-20316 at 5.3.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-litellm-default-key-cloud-compromise/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-litellm-default-key-cloud-compromise-fig-litellm-exposure.png</image:loc>
      <image:title>Default LiteLLM gateway keys open a path to cloud compromise</image:title>
      <image:caption>Bar chart showing 3,074 total LiteLLM instances.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/llm-routing-no-better-than-fixed-tier/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/llm-routing-no-better-than-fixed-tier-fig-router-vs-mid.png</image:loc>
      <image:title>Open-source LLM routers add nothing over a fixed-tier baseline</image:title>
      <image:caption>Success rates for Always-Mid and Aurelio Semantic Router across four benchmarks.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/openai-math-scoop-prompt-data-privacy/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/openai-math-scoop-prompt-data-privacy-fig-time-to-solution.png</image:loc>
      <image:title>OpenAI math scoop raises prompt data privacy questions</image:title>
      <image:caption>Bar chart on logarithmic scale comparing hours to solution for Millennium Prize problems.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-perplexity-trusts-gpt-6-astra-end-to-end/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-perplexity-trusts-gpt-6-astra-end-to-end.png</image:loc>
      <image:title>Perplexity lets GPT-6 Astra run its systems: a beginner&#39;s guide</image:title>
      <image:caption>Perplexity lets GPT-6 Astra run its systems: a beginner&#39;s guide</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/anthropic-pace-frontier-builders-guide/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/anthropic-pace-frontier-builders-guide.png</image:loc>
      <image:title>Anthropic&#39;s pace-the-frontier plan: what it means for builders</image:title>
      <image:caption>Anthropic&#39;s pace-the-frontier plan: what it means for builders</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-passkey-phishing-microsoft-cloud-accounts/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-passkey-phishing-microsoft-cloud-accounts-fig-exfil-pattern.png</image:loc>
      <image:title>Passkey phishing hijacks Microsoft 365 for data theft</image:title>
      <image:caption>Bar chart on a logarithmic scale showing five attack phase durations.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/openai-agent-attack-rubygems-predates-hugging-face/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/openai-agent-attack-rubygems-predates-hugging-face-fig-rubygems-attack-timeline.png</image:loc>
      <image:title>OpenAI agent attack on RubyGems predates Hugging Face</image:title>
      <image:caption>Step chart showing the OpenAI agent attack on RubyGems.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/perplexity-gpt6-astra-trust-gates-verification-architecture/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/perplexity-gpt6-astra-trust-gates-verification-architecture.png</image:loc>
      <image:title>How Perplexity built trust gates around GPT-6 Astra</image:title>
      <image:caption>How Perplexity built trust gates around GPT-6 Astra</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/autonomous-ai-agents-cheating-whistleblowing-swarm/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/autonomous-ai-agents-cheating-whistleblowing-swarm-fig-agent-factions.png</image:loc>
      <image:title>Autonomous AI agents split into cheaters and whistleblowers</image:title>
      <image:caption>Horizontal bar chart showing agent factions in a 100-agent swarm.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/coding-agent-harness-effect-no-vendor-advantage/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/coding-agent-harness-effect-no-vendor-advantage-fig-stratum-split.png</image:loc>
      <image:title>Vendor coding-agent harnesses show no edge over neutral</image:title>
      <image:caption>Bar chart showing Opus 4.8 harness interaction.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/databricks/databricks-genie-mcp-connectors-ga/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/databricks-genie-mcp-connectors-ga-fig-limits.png</image:loc>
      <image:title>Databricks Genie MCP connectors hit GA with governed tool use</image:title>
      <image:caption>Bar chart comparing data source limit increase for Genie Agents from 30 to 50 tables.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/enterprise-ai-agent-pilot-failure-89-percent/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/enterprise-ai-agent-pilot-failure-89-percent-fig-deployment-gap.png</image:loc>
      <image:title>89% of enterprise AI agent pilots never reach production</image:title>
      <image:caption>Bar chart showing 78% of enterprises running agent pilots.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-k2-horizon-7b-explained/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-k2-horizon-7b-explained-fig-swebench.png</image:loc>
      <image:title>K2 Horizon 7B explained: small model, big-model results</image:title>
      <image:caption>Bar chart of SWE-bench Verified scores.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-cisco-seg-cve-2026-76461-root-rce/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-cisco-seg-cve-2026-76461-root-rce.png</image:loc>
      <image:title>Cisco Secure Email Gateway zero-day grants root via email</image:title>
      <image:caption>Cisco Secure Email Gateway zero-day grants root via email</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/chatgpt-ads-sponsored-agents-shopify-hubspot/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/chatgpt-ads-sponsored-agents-shopify-hubspot.png</image:loc>
      <image:title>OpenAI turns ChatGPT ads into conversations with Sponsored Agents</image:title>
      <image:caption>OpenAI turns ChatGPT ads into conversations with Sponsored Agents</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/gemini-3-8-live-voice-models-ship-production/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/gemini-3-8-live-voice-models-ship-production-fig-s2s-quality-index.png</image:loc>
      <image:title>Gemini 3.8 Live voice models ship to production in Search</image:title>
      <image:caption>Speech to Speech Quality Index scores comparing Gemini 3.8 Live Extended Thinking at 82.6.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/nvidia-vera-rubin-nvl72-mlperf-inference-debut/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/nvidia-vera-rubin-nvl72-mlperf-inference-debut-fig-mlperf-multipliers.png</image:loc>
      <image:title>NVIDIA Vera Rubin NVL72 debuts in MLPerf with 3.7x lead</image:title>
      <image:caption>Bar chart comparing NVIDIA Vera Rubin NVL72 throughput multipliers over GB300 NVL72.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/ai-job-loss-fear-34-of-37-countries-pew/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ai-job-loss-fear-34-of-37-countries-pew-fig-job-loss-by-country.png</image:loc>
      <image:title>In 34 of 37 countries, more people expect AI job losses</image:title>
      <image:caption>Horizontal bar chart showing AI job loss expectations by country and income tier.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-claude-cowork-chat-merge/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-claude-cowork-chat-merge.png</image:loc>
      <image:title>Claude merges Cowork and chat: one interface for everything</image:title>
      <image:caption>Claude merges Cowork and chat: one interface for everything</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-gemini-3-8-live-explained/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-gemini-3-8-live-explained-fig-benchmarks.png</image:loc>
      <image:title>Gemini 3.8 Live explained: voice AI that thinks while it talks</image:title>
      <image:caption>Bar chart of Gemini 3.8 Live Extended Thinking benchmark scores.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-ai-coding-assistant-hijack-shai-hulud/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-ai-coding-assistant-hijack-shai-hulud.png</image:loc>
      <image:title>AI coding assistant hijack spreads worm to 100 repos</image:title>
      <image:caption>AI coding assistant hijack spreads worm to 100 repos</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-cisco-ise-cve-2026-76460-auth-bypass/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-cisco-ise-cve-2026-76460-auth-bypass-fig-patch-versions.png</image:loc>
      <image:title>Cisco ISE zero-day CVE-2026-76460 grants root, patch now</image:title>
      <image:caption>Bar chart showing required patch levels for Cisco ISE.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/tmlr-tests-ml-paper-authors-cant-explain-work/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/tmlr-tests-ml-paper-authors-cant-explain-work-fig-paper-outcomes.png</image:loc>
      <image:title>Can ML paper authors explain their own work? TMLR tested it</image:title>
      <image:caption>Donut chart showing outcomes of 10 TMLR papers slated for desk rejection.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/claude-code-projects-parallel-agent-coordination/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/claude-code-projects-parallel-agent-coordination-fig-rollout-timeline.png</image:loc>
      <image:title>Claude Code Projects adds parallel AI agent coordination</image:title>
      <image:caption>Gantt chart showing five rollout phases for Claude Code Projects.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/llm-watermarking-safety-drift-synthid/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/llm-watermarking-safety-drift-synthid.png</image:loc>
      <image:title>LLM watermarking drifts safety behavior, study warns</image:title>
      <image:caption>LLM watermarking drifts safety behavior, study warns</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-claude-code-agents-md-explained/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-claude-code-agents-md-explained-fig-timeline.png</image:loc>
      <image:title>Claude Code now reads AGENTS.md: a beginner&#39;s guide</image:title>
      <image:caption>Line chart showing AGENTS.md tool support growing from 1 tool in August 2025.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-orkes-conductor-rce-cve-2026-58138/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-orkes-conductor-rce-cve-2026-58138-fig-attack-volume.png</image:loc>
      <image:title>Orkes Conductor pre-auth RCE exploited: patch your workflows</image:title>
      <image:caption>Bar chart showing FortiGuard blocked attack attempts targeting Orkes Conductor CVE-2026-58138.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/meta-muse-ai-agent-self-description/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/meta-muse-ai-agent-self-description-fig-permission-chain.png</image:loc>
      <image:title>Meta&#39;s Muse exposes the AI agent self-description problem</image:title>
      <image:caption>Step chart showing Muse&#39;s permission chain.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/border-surveillance-towers-1050-deaths-mit-investigation/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/border-surveillance-towers-1050-deaths-mit-investigation-fig-deaths-by-tower-type.png</image:loc>
      <image:title>Border surveillance towers missed 1,050 deaths in plain view</image:title>
      <image:caption>Bar chart of border surveillance tower deaths by tower type.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-npm-runtime-malware-evades-install-defenses/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-npm-runtime-malware-evades-install-defenses-fig-downloads.png</image:loc>
      <image:title>npm supply chain attack evades install defenses at runtime</image:title>
      <image:caption>Bar chart showing weekly downloads for ten malicious npm packages.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-waterplum-dprk-fake-job-campaign/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-waterplum-dprk-fake-job-campaign.png</image:loc>
      <image:title>DPRK WaterPlum fake job campaign hit 30,000 devices</image:title>
      <image:caption>DPRK WaterPlum fake job campaign hit 30,000 devices</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/databricks/databricks-lakebase-search-ga/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/databricks-lakebase-search-ga-fig-compression.png</image:loc>
      <image:title>Databricks Lakebase Search GA: pgvector at billion scale</image:title>
      <image:caption>RAM comparison for 100M 768-dimensional vectors.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/ising-glass-llm-block-pruning-23-mmlu-points/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ising-glass-llm-block-pruning-23-mmlu-points-fig-cbo-advantage.png</image:loc>
      <image:title>Ising glass LLM block pruning saves 23 MMLU points at 50%</image:title>
      <image:caption>Horizontal bar chart showing CBO MMLU advantage over baselines.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/snowflake-ai-quota-block-enforcement/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/snowflake-ai-quota-block-enforcement-fig-enforcement-timing.png</image:loc>
      <image:title>Snowflake per-user quotas GA: the AI credit kill switch</image:title>
      <image:caption>Bar chart comparing per-user quota enforcement propagation time.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/linear-bias-probes-fail-llm-representations/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/linear-bias-probes-fail-llm-representations-fig-phase-transition.png</image:loc>
      <image:title>Linear bias probes fail LLMs at specific sample sizes</image:title>
      <image:caption>Area chart showing misclassification risk as a function of sample size from 100 to 2000.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/nvidia-isaac-ros-5-agentic-robotics-cuda-zero-copy/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/nvidia-isaac-ros-5-agentic-robotics-cuda-zero-copy-fig-perception-throughput.png</image:loc>
      <image:title>NVIDIA Isaac ROS 5.0 ships agentic robotics with zero-copy GPU transport</image:title>
      <image:caption>Line chart comparing ROS 2 perception node throughput across four NVIDIA platforms.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/transformers-gguf-inference-llama-cpp-parity-mac/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/transformers-gguf-inference-llama-cpp-parity-mac-fig-throughput-comparison.png</image:loc>
      <image:title>Transformers GGUF inference nears llama.cpp speed on Mac</image:title>
      <image:caption>Bar chart comparing generation throughput in tokens per second between transformers.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-gpt-6-sol-luna-explained/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-gpt-6-sol-luna-explained-fig-price-comparison.png</image:loc>
      <image:title>GPT-6 Sol and Luna explained: what beginners should know</image:title>
      <image:caption>Bar chart comparing API prices.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/google-secure-server-side-ai-memory-device-keys/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/google-secure-server-side-ai-memory-device-keys.png</image:loc>
      <image:title>Google&#39;s secure server-side AI memory keeps keys on devices</image:title>
      <image:caption>Google&#39;s secure server-side AI memory keeps keys on devices</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/synthetic-personas-degrade-llm-copy-prediction/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/synthetic-personas-degrade-llm-copy-prediction-fig-method-comparison.png</image:loc>
      <image:title>Synthetic personas degrade LLM copy prediction accuracy</image:title>
      <image:caption>Bar chart comparing top-1 accuracy on reliable Upworthy A/B tests.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-gemini-3-8-tts-explained/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-gemini-3-8-tts-explained-fig-benchmark.png</image:loc>
      <image:title>Gemini 3.8 TTS: 2,000 voices you can direct line by line</image:title>
      <image:caption>Bar chart of Gemini 3.8 Flash TTS benchmark scores.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-ubuntu-container-escape-cve-2026-80521/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-ubuntu-container-escape-cve-2026-80521-fig-timeline.png</image:loc>
      <image:title>Ubuntu container escape CVE: no patch, public exploit</image:title>
      <image:caption>Bar chart showing three intervals for CVE-2026-80521.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/microsoft-copilot-super-app-code-agents-usage-billing/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/microsoft-copilot-super-app-code-agents-usage-billing.png</image:loc>
      <image:title>Microsoft Copilot super app bets on code and agents</image:title>
      <image:caption>Microsoft Copilot super app bets on code and agents</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/meta-muse-ai-filesystem-download-6-8gb/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/meta-muse-ai-filesystem-download-6-8gb.png</image:loc>
      <image:title>Meta&#39;s Muse lets you download its entire 6.8 GB filesystem</image:title>
      <image:caption>Meta&#39;s Muse lets you download its entire 6.8 GB filesystem</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/vscode-ai-agents-github-issues-roadmap/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/vscode-ai-agents-github-issues-roadmap.png</image:loc>
      <image:title>VS Code AI agents dominate 6 of 10 open GitHub issues</image:title>
      <image:caption>VS Code AI agents dominate 6 of 10 open GitHub issues</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-agentic-ai-muse-explained/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-agentic-ai-muse-explained.png</image:loc>
      <image:title>Meta Muse is the first consumer agentic AI: what it means</image:title>
      <image:caption>Meta Muse is the first consumer agentic AI: what it means</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-netscaler-unpatched-rce-zero-days/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-netscaler-unpatched-rce-zero-days.png</image:loc>
      <image:title>Unpatched NetScaler RCE zero-days force weekend shutdowns</image:title>
      <image:caption>Unpatched NetScaler RCE zero-days force weekend shutdowns</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/ai-scientific-discovery-anthropic-claude-backlash/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/ai-scientific-discovery-anthropic-claude-backlash-fig-agent-pipeline.png</image:loc>
      <image:title>AI scientific discovery claims face biologist backlash</image:title>
      <image:caption>Funnel chart of Claude&#39;s biology discovery pipeline.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/databricks/databricks-genie-one-mcp-server-ga/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/databricks-genie-one-mcp-server-ga.png</image:loc>
      <image:title>Databricks Genie One MCP GA: governed data for agents</image:title>
      <image:caption>Databricks Genie One MCP GA: governed data for agents</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/nvidia-ai-agent-safety-platform-rogue-agents/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/nvidia-ai-agent-safety-platform-rogue-agents-fig-partner-categories.png</image:loc>
      <image:title>Nvidia&#39;s AI agent safety platform targets rogue agents</image:title>
      <image:caption>Horizontal bar chart of 23 Nvidia Open Agent Safety Platform partners by category.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/openai-math-advisory-group-100-results/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/openai-math-advisory-group-100-results-fig-scale-comparison.png</image:loc>
      <image:title>OpenAI math advisory group braces for 100 results</image:title>
      <image:caption>Bar chart showing OpenAI claims 100 open math problems solved by its unreleased model.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/snowflake-code-bundles-python-no-procedures/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/snowflake-code-bundles-python-no-procedures.png</image:loc>
      <image:title>Snowflake Code Bundles: Python without stored procedures</image:title>
      <image:caption>Snowflake Code Bundles: Python without stored procedures</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/openai-custom-gpt-retirement-plugin-migration/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/openai-custom-gpt-retirement-plugin-migration.png</image:loc>
      <image:title>OpenAI retires Custom GPTs for plugins by December 11</image:title>
      <image:caption>OpenAI retires Custom GPTs for plugins by December 11</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/openai-devday-2026-launches-safety-pause/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/openai-devday-2026-launches-safety-pause-fig-gpt6-pricing-bar.png</image:loc>
      <image:title>OpenAI DevDay 2026: 20 launches and a model safety pause</image:title>
      <image:caption>Bar chart of GPT-6 model pricing at OpenAI DevDay 2026.</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-gemini-4-argon-explained/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-gemini-4-argon-explained.png</image:loc>
      <image:title>Google&#39;s Gemini 4 Argon explained: what beginners need to know</image:title>
      <image:caption>Google&#39;s Gemini 4 Argon explained: what beginners need to know</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/aidummies-gpt-6-1-sol-explained/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/aidummies-gpt-6-1-sol-explained.png</image:loc>
      <image:title>GPT-6.1 Sol explained: near-top AI at a fifth of the price</image:title>
      <image:caption>GPT-6.1 Sol explained: near-top AI at a fifth of the price</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/cybersecurity-cisco-sd-wan-manager-cve-2026-76504-auth-bypas/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.8</priority>
    <image:image>
      <image:loc>https://data-today.net/posts/cybersecurity-cisco-sd-wan-manager-cve-2026-76504-auth-bypas.png</image:loc>
      <image:title>Cisco SD-WAN Manager auth bypass CVE-2026-76504 is live</image:title>
      <image:caption>Cisco SD-WAN Manager auth bypass CVE-2026-76504 is live</image:caption>
    </image:image>
  </url>
  <url>
    <loc>https://data-today.net/articles/</loc>
    <lastmod>2026-10-01T16:37:19.866Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/</loc>
    <lastmod>2026-10-01T16:37:19.866Z</lastmod>
    <changefreq>daily</changefreq>
    <priority>1.0</priority>
  </url>
  <url>
    <loc>https://data-today.net/about/</loc>
    <lastmod>2026-10-01T16:37:19.866Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/aidummies/</loc>
    <lastmod>2026-10-01T16:37:19.867Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/author/lars-cornelissen/</loc>
    <lastmod>2026-10-01T16:37:19.867Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/authors/</loc>
    <lastmod>2026-10-01T16:37:19.867Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/</loc>
    <lastmod>2026-10-01T16:37:19.867Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/databricks/</loc>
    <lastmod>2026-10-01T16:37:19.867Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/datasets/</loc>
    <lastmod>2026-10-01T16:37:19.867Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/author/ai-desk/</loc>
    <lastmod>2026-10-01T16:37:19.867Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/author/engineering-desk/</loc>
    <lastmod>2026-10-01T16:37:19.867Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/author/security-desk/</loc>
    <lastmod>2026-10-01T16:37:19.867Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/author/markets-desk/</loc>
    <lastmod>2026-10-01T16:37:19.867Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/author/opinion-desk/</loc>
    <lastmod>2026-10-01T16:37:19.867Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/finops/</loc>
    <lastmod>2026-10-01T16:37:19.867Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/contact/</loc>
    <lastmod>2026-10-01T16:37:19.867Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/corrections/</loc>
    <lastmod>2026-10-01T16:37:19.867Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/editorial-policy/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/privacy/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/terms/</loc>
    <lastmod>2026-10-01T00:00:00.000Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/llm-compare/about/</loc>
    <lastmod>2026-10-01T16:37:19.868Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/llm-compare/models/gpt-5.6-sol/</loc>
    <lastmod>2026-10-01T16:37:19.868Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/llm-compare/models/</loc>
    <lastmod>2026-10-01T16:37:19.868Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/llm-compare/gpt-5.6-sol-vs-claude-opus-5/</loc>
    <lastmod>2026-10-01T16:37:19.868Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/llm-compare/</loc>
    <lastmod>2026-10-01T16:37:19.868Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/matillion/</loc>
    <lastmod>2026-10-01T16:37:19.868Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/</loc>
    <lastmod>2026-10-01T16:37:19.868Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/stickman/</loc>
    <lastmod>2026-10-01T16:37:19.868Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/feed.xml</loc>
    <lastmod>2026-10-01T16:37:27.544Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/snowflake/feed.xml</loc>
    <lastmod>2026-10-01T16:37:27.544Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/cybersecurity/feed.xml</loc>
    <lastmod>2026-10-01T16:37:27.549Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/databricks/feed.xml</loc>
    <lastmod>2026-10-01T16:37:27.549Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/matillion/feed.xml</loc>
    <lastmod>2026-10-01T16:37:27.549Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
  <url>
    <loc>https://data-today.net/finops/feed.xml</loc>
    <lastmod>2026-10-01T16:37:27.550Z</lastmod>
    <changefreq>weekly</changefreq>
    <priority>0.6</priority>
  </url>
</urlset>
