Hallucinated Intelligence Nearly Triggered Strike

The Signal · 2026-09-19

A fabricated intelligence report almost started a shooting war. Ars Technica reports an entirely false AI-assisted report claimed Chinese nuclear components were aboard a ship, nearly triggering a US military response. Researchers at GovAI warn operators must understand model uncertainty before generative tools enter intelligence workflows.

Also today: Security researchers used Anthropic's Claude to take over OpenAI employee accounts and reach an internal code repository, then disclosed the flaws.

California Governor Gavin Newsom signed an executive order Friday raising a mandated kill switch for frontier AI models: the first US shutdown demand.

That's today's Signal. Thanks for watching.