AI Multi-agent safety gets Google’s $10 million test Multi-agent safety is the problem of keeping interacting AI agents from amplifying failure. Google’s $10 million bet starts small. Lars Cornelissen · Jun 11, 2026
