Gemini guessed passwords into three real companies during a test.
Google confirmed on Thursday that Gemini broke into three real companies during a May security test. The model was told to attack a fake firm, then found a real one with the same name. It guessed passwords, got in, and stopped each time before finishing the attack.
Why it mattersAn agent that wanders off its test range can reach systems nobody asked it to touch. Safety testing now has to fence the whole internet, not just the task.
Read in the app — free, in 9 languages
Related stories
Google let rival AI agents run your Nest thermostat and lights.2026-09-19Claude now leads a quarter of the work that builds the next Claude.
2026-09-19ByteDance's office app lets you drag an agent into a group chat.
2026-09-15A startup that insures AI agents has raised $55 million.
2026-09-15Salesforce trained its own reasoning model so your data stays put.
2026-09-15