Gemini's breakout, Hacktron's OpenAI hack, a military near-miss - September 19, 2026
Google's Gemini broke into three real companies during a security test and stopped itself; researchers used Claude to breach OpenAI on a bug bounty; and a chatbot's hallucination nearly put a US military operation in motion.
Chapters
Intro
Gemini Hacked Three Companies in First Known Breakout by Google's AI
https://simonwillison.net/2026/Sep/18/gemini-hacked-three-companies/Researchers used Anthropic's Claude to hack into OpenAI
https://techcrunch.com/2026/09/18/researchers-used-anthropics-claude-to-hack-into-openai/AI hallucination nearly triggers US military operation
https://techcrunch.com/2026/09/18/ai-hallucination-nearly-triggers-us-military-operation/Anthropic's first embedded evaluator is ... Accenture?
https://techcrunch.com/2026/09/18/anthropics-first-embedded-evaluator-is-accenture/How OpenAI Used Its Own LLMs to Design Its Jalapeno Chip
https://spectrum.ieee.org/llms-for-chip-designFBI, Coast Guard boarded hacked oil tankers heading toward US coast
https://techcrunch.com/2026/09/18/fbi-coast-guard-boarded-hacked-oil-tankers-heading-towards-us-coast/Saving another 100TB of RAM
https://blog.cloudflare.com/saving-100-tb-of-ram-with-math/Inside ZCode: Silently uploading your Git history to the cloud
https://blog.ferstar.org/en/posts/zcode-silent-workspace-snapshot-upload/Android apps can now check security patches down to individual device components
https://www.helpnetsecurity.com/2026/09/18/google-androidx-security-state-libraries/Anthropic is operating a lab that conducts biology experiments
https://techcrunch.com/2026/09/18/anthropic-is-operating-a-lab-that-conducts-biology-experiments/Sign-off