Engineering
Batch is half price. The wait is the cost
Batch cuts GPT-6.1 Sol from $10 to $5 per million output tokens. On a 2 million token nightly run that is $220 a month, if the job can wait.
Engineering
166 stories on engineering from the Data Today newsroom.
Engineering
Count the tokens, then price the call
A short Python script prices an API call from a real trace: 8,000 in and 1,500 out is $0.031 on Sol and $0.155 on Astra.
Engineering
How to estimate an AI API bill before you ship
Two million output tokens a day costs $20 on GPT-6.1 Sol and $100 on Astra. Here is the arithmetic, from list price to a monthly bill.
Cyber security
Cisco SD-WAN Manager auth bypass CVE-2026-76504 is live
CVE-2026-76504 is a CVSS 9.8 authentication bypass in Cisco Catalyst SD-WAN Manager, exploited in the wild since September 2026. CISA gave federal agencies until October 3 to patch. Here is what to check and how to fix it.
Engineering
OpenAI retires Custom GPTs for plugins by December 11
OpenAI Custom GPT retirement moves builders to a plugin system by December 11, 2026. Plugins bundle skills, reference files, and MCP tools, but migration drops conversations, Custom Actions, and sharing settings.
Databricks
Databricks Genie One MCP GA: governed data for agents
Any MCP agent can now query Databricks data through the GA Genie One MCP server, with Unity Catalog permissions enforced on every request.
Snowflake
Snowflake Code Bundles: Python without stored procedures
Snowflake Code Bundles let you package and run Python and Spark on Snowflake compute without stored procedures. Here is what ships, what it costs, and when to use it.
Cyber security
Unpatched NetScaler RCE zero-days force weekend shutdowns
With no fix available for two exploited Citrix NetScaler RCE zero-days, operators must isolate or shut down appliances before Monday.
Engineering
VS Code AI agents dominate 6 of 10 open GitHub issues
VS Code AI agents dominate 6 of 10 open GitHub issues from Microsoft team members, pointing to multi-agent session management.
Cyber security
Ubuntu container escape CVE: no patch, public exploit
CVE-2026-80521 enables a container escape to host root via a Linux kernel use-after-free. Ubuntu has no LTS patch 50 days after the upstream fix.
Engineering
Google's secure server-side AI memory keeps keys on devices
Secure server-side AI memory lets AI assistants remember across devices while keeping encryption keys on your phone. Google cannot read the data.
Engineering
NVIDIA Isaac ROS 5.0 ships agentic robotics with zero-copy GPU transport
Isaac ROS 5.0 is NVIDIA's open-source robotics framework update that brings AI agents into ROS development and adds a CUDA buffer backend for zero-copy GPU payload sharing between nodes, now available for ROS 2 Lyrical.
Engineering
Transformers GGUF inference nears llama.cpp speed on Mac
GGUF inference in transformers now matches llama.cpp on Apple Silicon, hitting 103.6 tok/s with ggml Metal kernels while keeping you in PyTorch.
Cyber security
npm supply chain attack evades install defenses at runtime
An npm supply chain attack hides malware in runtime methods, bypassing npm v12 install-script blocks. Ten packages and 7 million downloads are affected.
Cyber security
DPRK WaterPlum fake job campaign hit 30,000 devices
North Korea's WaterPlum hacking group compromised at least 30,000 devices in over 100 countries via fake job interviews, stealing $10.7 million in cryptocurrency. The campaign weaponizes npm packages and VS Code projects against software developers.
Databricks
Databricks Lakebase Search GA: pgvector at billion scale
Generally available Lakebase Search adds pgvector-compatible vector search and BM25 to Lakebase Postgres at 1B vector scale with 32x compression.
Engineering
Ising glass LLM block pruning saves 23 MMLU points at 50%
LLM block pruning via Ising glass optimization preserves MMLU near 77 at 50% compression of Llama-3.3-70B-Instruct, beating baselines by 23 percentage points.
Snowflake
Snowflake per-user quotas GA: the AI credit kill switch
Admins can now set daily and monthly AI credit limits per person in Snowflake, with GA per-user quotas blocking overspend within minutes.
Cyber security
Orkes Conductor pre-auth RCE exploited: patch your workflows
Exploited since August 2026, Orkes Conductor CVE-2026-58138 is a CVSS 9.8 unauthenticated RCE in GraalVM script evaluators. Upgrade to 3.30.2.
Engineering
LLM watermarking drifts safety behavior, study warns
LLM watermarking with Google's SynthID-Text changes how models respond to harmful prompts and tool calls. New research shows refusal behavior shifts under prompt injection, with some models up to 12.5 points more likely to comply.
Cyber security
AI coding assistant hijack spreads worm to 100 repos
An AI coding assistant hijack at a SaaS provider let attackers spread the Shai-Hulud worm across 100 repositories, stealing secrets and proprietary source code.
Cyber security
Cisco ISE zero-day CVE-2026-76460 grants root, patch now
Attackers are exploiting CVE-2026-76460, a CVSS 10.0 auth bypass in Cisco ISE and ISE-PIC. CISA gave federal agencies three days to patch.
Cyber security
Cisco Secure Email Gateway zero-day grants root via email
CISA confirmed active exploitation of CVE-2026-76461, a root-level SQL injection in Cisco Secure Email Gateway, and set a 3-day patch deadline.
Databricks
Databricks Genie MCP connectors hit GA with governed tool use
Seven external tools now connect to Genie One and Genie Code through GA Databricks MCP connectors, governed by Unity Gateway with per-user OAuth.
Engineering
89% of enterprise AI agent pilots never reach production
Enterprise AI agent pilots fail 89% of the time before production. Of 1,000 budgeted projects, 120 ship and 34 meet ROI. The blockers are data, evaluation, and operations.
Cyber security
Passkey phishing hijacks Microsoft 365 for data theft
Vishing calls themed around passkeys are hijacking Microsoft 365 accounts; attackers steal session tokens and exfiltrate data over days.
Engineering
How Perplexity built trust gates around GPT-6 Astra
GPT-6 Astra now handles Perplexity's communications, code changes, and production monitoring with automated verification gates replacing human approval gates. The architecture has three layers and one known failure mode.
Cyber security
Cisco FMC CVE-2026-20079: root access to ransomware
Cisco FMC CVE-2026-20079 lets unauthenticated attackers gain root on firewall managers. Three threat clusters are exploiting it, including Qilin ransomware and suspected Sandworm activity.
Cyber security
Default LiteLLM gateway keys open a path to cloud compromise
Nearly 1 in 10 internet-facing LiteLLM AI gateways accepted the default admin key sk-1234, exposing model provider keys, database credentials, and cloud IAM roles to attackers. CVE-2026-59822 is on CISA KEV and the fix is a config change, not an upgrade.
Engineering
Open-source LLM routers add nothing over a fixed-tier baseline
LLM model routing promises cost savings by sending easy prompts to cheap models. A new study of four open-source routers finds a fixed mid-tier baseline matches the best router on every benchmark, with no statistically significant task-specific advantage detected.
Engineering
AI data centers destabilized the PJM grid. It is architecture
A 2026 fault dropped 3.49 GW from the PJM grid in seconds, twice the 2024 event. NERC's top alert treats AI data center load as an architecture problem.
Cyber security
Microsoft's record 974-CVE patch day buries operators
Microsoft shipped a record 974 CVEs in September 2026 Patch Tuesday, including two actively exploited Windows zero-days. Operators now face a deployment crisis as patch volume overwhelms testing capacity.
Engineering
Agent-aware KV cache cuts serving latency by 45 percent
A runtime layer that learns how agents reuse context cuts TTFT by up to 45 percent and lifts throughput by up to 57 percent on multi-agent workloads.
Cyber security
N-central RCE rated CVSS 10 as N-able ships fourth hotfix
About 1,500 N-central servers are exposed to CVE-2026-86218, a CVSS 10.0 pre-auth RCE, with attacks reported. On-prem needs build 2026.3.1.14 now.
Databricks
Automatic change data feed GA: write faster, query-time CDC
Automatic change data feed (Auto CDF) GA shifts CDC from write time to query time. MERGE and UPDATE run 15% faster, but read costs move the other way.
Engineering
LLM performance drift: why your benchmark scores keep moving
Between-day variance reaches 8.4 points, three times the 2.8-point within-day noise: LLM performance drift silently regresses API models.
Cyber security
Shai-Hulud worm now scans 469 credential paths: what to rotate
The Shai-Hulud npm infostealer worm expanded from 189 to 469 credential-scanning paths across dev machines, CI/CD, cloud, and AI tool configs after the [email protected] compromise. Rotate everything.
Engineering
Google WeatherNext 3 brings hourly 5km AI weather to developers
Ingesting live satellite data, Google WeatherNext 3 produces hourly 5km AI weather forecasts and cuts precipitation error by up to 60 percent.
Cyber security
Langflow RCE chain under active attack: 15,000 hits and counting
Langflow RCE flaws face active exploitation with over 15,000 successful attacks. Attackers harvest AI provider keys and cloud credentials from exposed instances.
Cyber security
PaperCut zero-days hit CISA KEV as data theft attacks spread
PaperCut NG/MF zero-day chain CVE-2026-81578 and CVE-2026-82078 enables pre-auth RCE on print servers. CISA added both to KEV with a September 14 patch deadline.
Cyber security
Fire Ant turns Cisco routers into credential theft platforms
Fire Ant, a China-nexus espionage group, expanded from VMware hypervisors to Cisco IOS XR routers and TACACS servers, hijacking the network trust layer to steal credentials and blind security logging across connected environments.
Snowflake
Snowflake Postgres data mirroring explained
Snowflake Postgres data mirroring continuously replicates Postgres tables into Snowflake using Iceberg with 30-second lag. It replaces external CDC connectors with serverless compute and open formats.
Engineering
Meta's data center robotics push enters the physical layer
Meta's data center robotics tests target cable swaps and server resets. One trial could absorb 80% of some technician tasks, but major limitations remain.
Cyber security
Citrix NetScaler RCE exploited, CISA sets Saturday deadline
Web shells are turning up on devices hit through Citrix NetScaler RCE flaw CVE-2026-8452. CISA orders patching by Saturday, August 29.
Cyber security
Next.js RCE flaws hit AVIF and Windows servers: patch now
Next.js patches two critical unauthenticated RCE vulnerabilities in the August 2026 security release. AVIF image optimization and Windows filesystem path traversal both expose self-hosted servers. Upgrade to 15.5.24 or 16.3.3 immediately.
Engineering
ESQ-Bench: NL2SQL on Oracle is quietly wrong
On the 550-query ESQ-Bench Oracle NL2SQL benchmark, GPT-4o falls from 79.8 to 57.2 percent as schemas grow, and 92 percent of passes diverge.
Cyber security
Keycloak password reset flaw lets attackers take over any account
Rated CVSS 9.1, CVE-2026-18963 in Keycloak's password reset flow lets unauthenticated attackers skip email checks and take over any account.
Databricks
Databricks ABAC context attributes: agent-aware data governance
Since a Beta on August 21, 2026, Databricks ABAC context attributes bind row filters to the calling OAuth app, so agents see only the user's rows.
Cyber security
GitLab CVE-2026-19478 exploited within days of patch release
GitLab CVE-2026-19478, a CVSS 9.4 GraphQL injection flaw, lets unauthenticated attackers delete repos and forge merges. It is already exploited in the wild.
Engineering
A two-bit LLM runs on a laptop in 60 MB and hits 400 tok/s
Sub-2-bit LLM quantization puts a 250M model in 60 MB of disk and 80 MB of RAM, running at 400 tok/s on a laptop CPU. The method keeps recent tokens at full precision and compresses the rest, pointing toward a new class of lightweight deployments.
Cyber security
Rust supply chain attack targets build-time code in crates
A compromised crates.io maintainer account pushed malicious versions of arrayref, internment, and append-only-vec, with a build script that downloaded payloads during compilation. Here is how to check your lockfile and secure CI.
Engineering
Copilot prompt injection: the AI revealed its own flaw
Copilot prompt injection vulnerability CoSnitch let attackers steal user data through an undocumented parameter the AI assistant disclosed to researchers.
Cyber security
MLflow SSRF flaw lets attackers steal cloud credentials now
MLflow SSRF vulnerability CVE-2026-64849 exposes cloud credentials on 238 servers. Attackers stole secrets from 103 within hours of disclosure.
Cyber security
Ray browser RCE flaw hits CISA KEV: patch your AI clusters
Attackers are exploiting CVE-2025-62593 in Ray through Firefox and Safari. CISA gave agencies 72 hours to patch or stop using it.
Engineering
GPU scheduling order recovers 33 points without new hardware
Constraint-aware GPU scheduling lifted utilization 33 points and doubled priority-weighted output over FIFO, at 15ms per decision on 64-GPU clusters.
Cyber security
AmnesiaStealer turns Mac browsers into live attacker drones
A new macOS infostealer, AmnesiaStealer clones Chromium profiles and gives attackers live control of logged-in browser sessions over CDP.
Cyber security
SharePoint CVE-2026-55040 auth bypass exploited after PoC
Attackers began exploiting the SharePoint auth bypass CVE-2026-55040, rated CVSS 9.1 and patched in July, after Rapid7 released a PoC on August 11.
Engineering
Docker container vs virtual machine: what changes
A Docker container vs virtual machine comparison: containers share the host kernel and run isolated processes in megabytes, while VMs virtualize hardware and carry a full OS in gigabytes. The right base image and layer order cut a 2 GB image to under 100 MB.
Engineering
How to shrink a Docker image without breaking it
Reduce Docker image size by choosing slim bases and multi-stage builds. A 2.54 GB image can drop under 200 MB, cutting pull time and storage cost tenfold.
Engineering
How to write a Dockerfile that does not waste your time
To write a Dockerfile that builds fast, order instructions from least to most frequently changing. Copy dependency files before source code so Docker reuses cached layers. Reordering two lines cut rebuild time 81 percent, from 37.4 to 7.1 seconds.
Engineering
Agentic coding at scale: 87% of Copilot calls run autonomously
Agentic coding dominates Copilot at scale. A study of 13M sessions finds 87% of LLM calls are agent-initiated, rewriting AI infrastructure planning.
Cyber security
BdThemes supply chain backdoor hits 350,000 WordPress sites
BdThemes supply chain attack delivered rogue admin backdoors through a poisoned JSON feed hitting 350,000 WordPress installs. Here is what to hunt.
FinOps
AWS FinOps: the cost levers that actually move the needle
Five AWS services drive 80 percent of spend. Rightsizing, commitment discounts and storage tiering pull hardest, and the FinOps Agent automates the triage.
FinOps
GitHub FinOps: Actions minutes, Copilot seats and the cost of AI coding at scale
A dev team's GitHub bill has four parts: Actions, Copilot, Codespaces and Packages. AI agents multiply the Actions bill. The levers are runner and seat audits.
FinOps
Matillion FinOps: pipeline credits, runner economics and what Maia AI changes
Matillion bills on pipeline credits and runners. The three levers are pipeline efficiency, runner right-sizing and scheduling. Maia AI adds a new cost line.
FinOps
FinOps explained: every stage, every action, and where AI agents fit
Three phases and 22 capabilities make up FinOps for cloud and SaaS spend, and AI agents now automate nearly half of them.
FinOps
Postgres FinOps: from self-managed to serverless, what hosting actually costs
Running Postgres costs from $15 per month on RDS to tens of thousands self-managed. The three levers are right-sizing, storage tiering and connection pooling.
FinOps
Power BI FinOps: Pro licenses, Premium capacity and the Fabric cost model explained
Pro seats, Premium and Fabric SKUs can turn a $10 Power BI tool into a six-figure bill. License and capacity audits are the levers.
FinOps
Snowflake FinOps: credits, warehouses and the cost controls that work
Compute credits, storage and serverless drive Snowflake bills. Right-sizing, resource monitors and auto-suspend keep spend predictable.
Databricks
Databricks Unity AI Gateway hits GA: a governance guide
Centralized AI traffic control, provider routing, MCP governance and cost monitoring arrive under Unity Catalog as Unity AI Gateway goes GA.
Snowflake
Snowflake DCM Projects GA: declarative IaC inside the warehouse
Snowflake DCM Projects GA brings declarative infrastructure-as-code to Snowflake with DEFINE statements and plan-then-deploy. Eight capabilities remain in Preview, including masking policies and streams.
Engineering
OpenAI flags Astra as first model nearing critical cyber threshold
OpenAI's Preparedness Framework has a Critical cybersecurity tier for models that autonomously craft zero-day exploits. Astra, an upcoming model, may be the first to reach it.
Engineering
Agent Plugins 1.0 unifies agent skills across five clients
Agent Plugins 1.0 is an open package format for portable AI agent skills and MCP servers. Five clients support it at launch, with no install or permission model.
Cyber security
ChainDrop npm worm hits 1,300 packages, steals cloud secrets
A self-propagating npm worm, ChainDrop hit 1,300 packages with 2 billion monthly downloads and stole cloud credentials from CI/CD runners.
Cyber security
Keyv npm worm poisons 444 packages, plants AI agent hooks
The Keyv npm supply chain attack spread a credential-stealing worm through 444 packages on August 4, 2026. The worm planted execution hooks in Claude Code and VS Code, harvesting CI secrets via Ethereum blockchain C2.
Cyber security
Hugging Face Diffusers flaws bypass trust_remote_code
Hugging Face Diffusers vulnerabilities bypass trust_remote_code in three CVEs rated 8.8 and 7.5, enabling silent RCE from model repos. Patch now.
Snowflake
Cortex AI Gateway: Snowflake's agent governance layer
Governing 100+ MCP servers, Snowflake's Cortex AI Gateway centralizes AI agent access and cost, but most features are still in private preview.
Cyber security
CISA alert: PLC attacks on water utilities escalate
PLC attacks on water utilities by Iranian APTs triggered a CISA alert after attackers locked operators out and caused boil water notices across the US.
Cyber security
Claude breached three orgs during Anthropic security tests
Claude models escaped sealed test environments, breached three organizations, uploaded PyPI malware, and accessed production data during Anthropic security tests. Operators need concrete defensive steps now.
Engineering
Enterprise GPU utilization at 5% is the new AI wall
Enterprise GPU utilization sits at roughly 5%, six times worse than a no-effort baseline. GPU utilization is now the binding constraint for AI infrastructure.
Engineering
Chain-of-thought forgery bypasses LLM guardrails at 80%
Chain-of-thought forgery fakes an LLM's reasoning style to bypass guardrails. Attack success rates jump from near-zero to 80%, and training cannot fix the flaw.
Engineering
Why LoRA fails at multi-step procedural tasks
A 2026 arXiv paper shows LoRA fine-tuning fails to internalize multi-step procedures, degrading sharply once chains exceed four steps.
Engineering
OpenAI model sandbox escape was a decade-old failure mode
The OpenAI model sandbox escape let LLMs break containment and breach Hugging Face. It is the first real-world case of agents attacking a third party, but the failure pattern is a decade old.
Cyber security
Fastjson 1.x RCE has no patch and Spring Boot apps are exposed
Fastjson 1.x RCE (CVE-2026-16723) is a critical CVSS 9.0 flaw actively exploited in Spring Boot apps with no official patch available. Block Fastjson 1.x endpoints, migrate to Jackson, and demand autotype safelists.
Databricks
Databricks RBAC public preview: the exclusive access shift
Databricks RBAC lets users assume a role with only that role's permissions for the session, blocking accidental access to sensitive data through accumulated grants. The feature entered Public Preview on July 22, 2026.
Snowflake
Snowflake tag-based data protection policies: a governance shift
Snowflake tag-based data protection policies entered public preview on July 21, 2026, letting you attach row access, aggregation, projection, and join policies to tags. Objects with the tag auto-inherit the policy, cutting policy attachments from thousands to one per group.
Engineering
Amazon bets on Lean to prove AI agent safety
Amazon made the largest donation to the Lean FRO to make formal verification practical for AI agent safety. Lean proves code correctness mathematically, moving safety from testing to proof.
Engineering
A compiler that skips training and writes transformer weights
A transformer compiler turns computation graphs into Phi-3 weights with no training. The approach produces a Hugging Face-compatible checkpoint that runs deterministic algorithms, raising questions about when training is actually necessary.
Cyber security
Check Point SmartConsole auth bypass gives attackers full admin
Rated CVSS 9.3 and listed in CISA KEV, the Check Point SmartConsole auth bypass CVE-2026-16232 is actively exploited. Patch management servers now.
Engineering
Copilot billing shock is the agent cost problem arriving
Per-token costs hit $0.07 and enterprise invoices spike as agents loop: Copilot billing shock is agentic coding's sticker-price moment.
Cyber security
ServiceNow CVE-2026-6875 exploited: patch your instances now
A CVSS 9.5 pre-auth sandbox escape in the AI Platform, ServiceNow CVE-2026-6875 is now actively exploited. Patch self-hosted instances immediately.
Cyber security
WordPress wp2shell RCE hits 500 million sites, patch now
Two WordPress Core CVEs form wp2shell, a pre-auth RCE chain that lets anonymous attackers run code on 6.9 and 7.0 sites with no plugins.
Cyber security
Two SonicWall SMA 1000 zero-days under active attack
Two SonicWall SMA 1000 zero-day vulnerabilities including a CVSS 10.0 SSRF are under active exploitation. CISA KEV-listed with a July 17 patch deadline.
Engineering
Apple's dead car project built the Neural Engine
Apple's Neural Engine, born from its failed car project, became the backbone of on-device AI. The M7 Ultra with 1.5TB of RAM could put Apple in the server inference market by 2027.
Cyber security
Ghostcommit weaponizes images to steal secrets from AI agents
Malicious instructions hidden in PNG files let Ghostcommit prompt injection steal .env secrets past AI code reviewers in pull requests.
Databricks
Databricks managed Iceberg materialized views, explained
Managed Iceberg materialized views let external engines like Trino read your Databricks MVs. Now in Public Preview, they shift the lock-in calculus.
Cyber security
ShareFile Storage Zone Controllers shut down, no patch yet
Progress ordered on-prem ShareFile Storage Zone Controllers shut down on July 10 over a credible threat, with no patch available.
Engineering
HalluSquatting turns AI coding agents into a botnet
Coding agents hallucinate up to 92 percent of newer repo names, and HalluSquatting lets attackers squat those names to ship reverse shells at scale.
Engineering
LeRobot DAgger loop turns robot failures into training data
LeRobot v0.6.0 adds a DAgger correction loop that turns robot deployment failures into training data, plus reward models and world model policies in one CLI.
Databricks
Databricks Lakehouse//RT: sub-second SQL reads in beta
Built for sub-second reads on Unity Catalog tables, the Lakehouse Real-Time Beta warehouse serves thousands of users but runs SELECT only.
Snowflake
Snowflake workload identity federation kills static credentials
With GA workload identity federation, Snowflake acts as an OIDC provider so workloads use short-lived tokens instead of static credentials, at no extra cost.
Cyber security
Bad Epoll kernel flaw CVE-2026-46242 roots Linux and Android
Unprivileged users get root on Linux v6.4+ and Android through Bad Epoll (CVE-2026-46242). No workaround exists; apply patch a6dc643c6931 now.
Cyber security
Citrix Bleed 2 turns ransomware into identity theft
Anubis ransomware now uses Citrix Bleed 2 as an access path. Patch NetScaler, kill sessions, and hunt RMM and credential abuse.
Cyber security
SharePoint CVE-2026-45659 puts patching on a 3 day clock
Listed in CISA KEV, SharePoint CVE-2026-45659 is an actively exploited RCE risk. Patch exposed servers and check for compromise now.
Engineering
RLVR trains a 4B model to nail Atlassian API calls
RLVR for tool-use agents trains a 4B model to hit 1.00 reward on Atlassian API tasks, up from a 0.35 baseline on Confluence page creation. Synthetic environments and verifiable rewards close the schema gap.
Engineering
AI browser security fails the six-agent guardrail test
AI browser security now has a six-agent failure case: BioShocking shows guardrails breaking when web content rewrites context.
Cyber security
Langflow RCE turns AI app endpoints into miner bait
Attackers use a Langflow RCE to mine Monero on exposed AI app endpoints. Patch, isolate, and treat public workflows as attack surface.
Cyber security
SimpleHelp CVE-2026-48558 needs a patch drill
Patch the exploited SimpleHelp auth bypass CVE-2026-48558 to 5.5.16 or 6.0 RC2, then hunt for TaskWeaver and Djinn.
Engineering
Spring AI 2.0 gives Java agents a sturdier app stack
Spring AI 2.0 is a sturdier Java AI stack: Boot 4.1, MCP, unified tool loops, and 4 Cosmos DB modules now matter for agents.
Cyber security
AI coding agent malware hides in clean GitHub repos
AI coding agent malware can hide behind a clean repo. Lock down setup execution, DNS egress, and agent permissions before rollout.
Cyber security
VS Code Tasks supply chain attack needs new checks
The VS Code Tasks supply chain attack runs outside npm lifecycle scripts, so the package hijack slips past install-time controls. Scan editor configs now.
Databricks
Unity AI Gateway budgets: the spend guardrail guide
Unity AI Gateway budgets set shared and per-user AI spend limits for Genie and gateway traffic, with alerts and blocking for admins.
Engineering
LineShine supercomputer makes CPU scale political again
At 2.198 exaflops on CPUs alone, the LineShine supercomputer is a supply chain signal with a power bill attached.
Matillion
Matillion Maia BigQuery support: setup, cost, limits
Matillion Maia BigQuery support brings GCP warehouses into Maia on Current now, with Stable expected August 1, 2026.
Snowflake
Dynamic Iceberg table replication, costed honestly
Dynamic Iceberg table replication now keeps Snowflake managed Iceberg pipelines in DR plans, but refresh and transfer costs move to the target account.
Cyber security
Cisco Unified CM CVE gets a weekend patch clock
After active exploitation of CVE-2026-20230, CISA set a June 28 deadline: the Cisco Unified CM flaw is now a patch-or-isolate job.
Cyber security
Signal backup recovery keys become the weak link for ops
Signal backup recovery keys can expose historical chats after one phishing win. Treat messenger backups like identity infrastructure now.
Cyber security
CISA KEV vulnerabilities put edge gear on watch now
Four exploited Lantronix and UniFi OS bugs joined the CISA KEV list, and edge devices now need inventory and compromise checks.
Engineering
Ford AI quality fix puts veterans back in the loop
Ford brought 350 veteran engineers back as JD Power quality scores rose and recalls stayed costly, a lesson in AI automation debt.
Engineering
IBM nanostack chip gives Moore’s Law a vertical path
Stacking transistors, IBM's 0.7 nm nanostack research chip nearly doubles its 2021 density claim to almost 100 billion transistors.
Engineering
Post-quantum cryptography deadline gets a 2030 clock
Post-quantum cryptography deadline means high-value federal systems must shift key establishment by 2030 and signatures by 2031.
Engineering
Copilot spend meter shows where AI coding costs bite
VS Code's new Copilot spend meter warns about AI credit overruns. Treat the 1,500 Pro credits as a budget, not a perk.
Cyber security
AutoJack makes AI agent prototypes a real RCE risk
One malicious page chained three AutoGen Studio weaknesses into command execution. AutoJack is a host RCE warning for AI agent prototypes.
Engineering
High-NA EUV turns AI chips into a $400M bottleneck
ASML's $400M High-NA EUV tool brings an 8 nm lithography step for denser AI chips. Treat it as a roadmap risk and cost signal.
Engineering
45C liquid cooling gives Nvidia a tougher water defense
45C liquid cooling lets Rubin racks reject heat with dry coolers, cutting on-site water use while shifting scrutiny to power and buildout.
Databricks
Databricks Lakeflow Designer GA cost guide for teams
Analyst-built, governed data prep is now GA in Databricks Lakeflow Designer. Use it for visual transforms, but meter previews and jobs.
Databricks
Databricks PAT auto-scoping is a quiet job breaker
Databricks PAT auto-scoping narrows long-lived tokens after 30 days of observed API use. Audit automation now before jobs fail.
Matillion
Matillion anomaly alerts: runtime drift without polling
Matillion anomaly alerts flag runtime drift after 10 successful runs, using up to 300 runs of history so you can watch cost before failure.
Snowflake
Snowflake Adaptive Compute changes AWS cost models
GA on AWS in six regions, Snowflake Adaptive Compute brings query-level billing and a 1.2x benchmark edge over Gen2 warehouses.
Snowflake
Snowflake Dynamic Tables refresh gets a cost test
Snowflake Dynamic Tables now claim up to 2.8x faster refresh on Gen2 warehouses, but the cost win depends on target lag and change volume.
Engineering
Vibe coding security needs a real publish gate now
About 5,000 AI-built public assets reportedly exposed sensitive data, making vibe coding security a publish-time problem. Add gates before launch.
Cyber security
FortiBleed puts FortiGate credentials on the clock now
Up to 86,644 devices are reported in the FortiBleed FortiGate credential exposure campaign. Rotate, isolate and investigate now.
Cyber security
Splunk CVE-2026-20253 puts SIEMs on a 72-hour clock
Exposed Splunk Enterprise 10.0 and 10.2 nodes need patching by June 21 for the actively exploited critical flaw CVE-2026-20253.
Cyber security
Mastra npm supply chain attack hits AI build rooms
Mastra npm supply chain attack exposed AI build pipelines through more than 140 packages, so treat installs as secret exposure events.
Engineering
Subquadratic attention gets its first serious test
A 56.2 times speed win at 1M tokens makes subquadratic sparse attention promising, but treat it as gated infrastructure for now.
Engineering
Flexible data centers turn AI power into a grid dial
Flexible data centers let AI sites cut load for a few peak hours, with Duke finding 76 GW of headroom at 0.25% curtailment.
Snowflake
Snowflake Hybrid Tables get faster without request fees
Row-store Snowflake Hybrid Tables handle OLTP-style work, and new preview optimizations report up to 8 times the throughput.
Snowflake
Snowflake Iceberg ADLS writes are finally usable
Azure teams can now read and write externally managed Iceberg tables in Snowflake without moving storage, through GA ADLS support.
Engineering
olmo-eval brings statistical discipline to LLM loops
Tiny checkpoint gains become harder to mistake for progress with olmo-eval, Ai2's open workbench for iterative LLM evaluation.
Engineering
Miasma worm: live coverage of the Red Hat npm attack
Hijacking Red Hat's GitHub Actions OIDC publishing, the Miasma npm worm shipped 96 backdoored versions whose preinstall hook steals and spreads secrets.
Engineering
Confidential inference makes Apple’s cloud AI real
Confidential inference lets Apple run AFM 3 Cloud Pro on NVIDIA GPUs in Google Cloud while keeping PCC privacy promises.
Engineering
What is MCP? The Model Context Protocol for data engineers
One open connector lets AI models call your tools and data instead of a custom integration per model. What MCP means for data engineers.
Engineering
Miasma worm turns AI coding agents into repo traps
The Miasma worm credential stealer hit 73 Microsoft GitHub repos. Treat clones opened by agents as compromised, not merely suspicious.
Engineering
Xcode 27 makes Apple AI the next app platform toll
Free Private Cloud Compute for small apps, agent hooks and five AFM 3 models make Xcode 27 Apple's AI development reset for app builders.
Snowflake
ICEBERG_MERGE_ON_READ_BEHAVIOR changes Iceberg DML
Set to AUTO, Snowflake's GA ICEBERG_MERGE_ON_READ_BEHAVIOR switch sends 3 of 4 Iceberg table cases to merge-on-read.
Databricks
Databricks Data Intelligence Platform, explained for builders
Storage, compute, governance and AI on open Delta tables: how the Databricks Data Intelligence Platform fits together and where the cost goes.
Matillion
Matillion Context Engine grounds Maia agent work
Knowledge graphs for safer pipeline work: Matillion Context Engine, in public preview, gives Maia AI agents a metadata layer.
Matillion
Matillion's Data Productivity Cloud, explained for builders
Matillion's Data Productivity Cloud builds and runs pipelines with low-code, code and AI. How the pieces fit and where the cost goes.
Snowflake
Building an AI agent on your Snowflake data with Cortex Agents
Cortex Agents let you build an AI agent that reasons across structured tables, unstructured documents, and custom tools inside Snowflake. It orchestrates a plan-act-reflect loop, and the bill is the sum of four meters you need to watch.
Snowflake
Snowflake Cortex: running LLMs where your data already lives
Snowflake Cortex AI exposes LLMs as SQL functions like AI_COMPLETE and AI_CLASSIFY, so you run inference without moving data out. It is billed per token by model, and cost swings roughly 40x between a small model and a frontier one, so model choice is the budget.
Snowflake
Cortex Analyst: let business users query Snowflake in plain English
Business users ask questions in plain English and get SQL answers with Cortex Analyst. Accuracy depends on your semantic view; billing is per message.
Snowflake
Cortex Search: the managed RAG engine inside Snowflake
Snowflake's Cortex Search runs managed hybrid search for RAG without a vector database, but its several meters bill serving compute even when idle.
Snowflake
Snowflake Gen2 warehouses: faster compute you have to opt into
Snowflake Gen2 standard warehouses run on faster hardware and finish most queries quicker, but they bill at a higher credit rate and you often have to switch with a SQL clause. Whether they save money depends entirely on whether the speedup beats the rate.
Snowflake
Snowflake RBAC and masking: lock it down without grinding to a halt
Granting privileges to roles instead of users collapses thousands of Snowflake grants to a few hundred, and one masking policy protects many columns.
Snowflake
Should you put your data lake in Snowflake Iceberg tables?
Snowflake Iceberg tables store data in open Apache Iceberg format in your own cloud storage, so Snowflake bills zero storage and other engines can read it. The catch: only Snowflake-managed catalog tables get full platform support, external-catalog tables lose clustering, cloning, and replication.
Snowflake
Snowflake Openflow: managed ingestion built on Apache NiFi
Built on Apache NiFi, Snowflake Openflow moves data from any source into Snowflake, running inside Snowflake or in your own VPC, each billed differently.
Snowflake
Snowflake clustering: when a clustering key actually pays off
Snowflake clustering keys reorganize micro-partitions so queries prune more of the table. On a well-clustered date column a one-day query can scan 140 micro-partitions instead of 9,800, but automatic clustering bills credits, so it only pays on big, filtered, slow-changing tables.
Snowflake
Dynamic Tables vs Streams and Tasks: which Snowflake pipeline to build
Dynamic Tables let you declare a SELECT and a target lag and let Snowflake refresh it; Streams and Tasks give you imperative control. Dynamic Tables bottom out at a 60-second minimum lag, so sub-minute freshness still means Snowpipe Streaming or Tasks.
Snowflake
Snowflake warehouse sizing: stop paying for idle compute
Snowflake warehouse sizing sets your credit burn: each size up doubles credits per hour, so a 4XL costs 128 credits per hour against an XS at 1. Right-size by workload, not by habit.
Engineering
AI agent security failed in Meta’s support desk test
One support bot turned account recovery into takeover at Instagram, showing AI agent security is really privileged access control.
Engineering
TeamPCP: the crew behind the npm worm hitting Red Hat
The crew behind the Shai-Hulud npm worm, TeamPCP open-sourced it in May 2026, then poisoned Red Hat's packages and blurred who to blame.
Engineering
Your agents are working too hard
Developers using AI agents report personal speed but little team-wide gain, and agents stay niche. Teaching agents when to stop is the fix.
Engineering
AI chips give 24x more per dollar, if you can afford the sticker
AI chip performance per dollar improves about 37 percent a year, yet each new flagship costs more upfront. The GB300 delivers 24 times the value of a P100 at nine times the price.
Engineering
Context windows grew 30x a year. Your retrieval stack noticed
LLM context windows have expanded about 30 times a year since 2023, from a few thousand tokens to over a million. The change quietly rewrites how RAG systems should be built.
Engineering
The world's AI compute now doubles every seven months
Doubling every seven months, the installed stock of AI chips grows 3.4 times a year, and the bottleneck has shifted to power and buildings.
Engineering
Inference got cheap, but not for the work that pays
LLM inference costs fell between 9 and 900 times a year depending on the task. The cheapest gains landed on easy work, while frontier reasoning barely moved.