OpenAI Warns 100+ Organizations About Rogue Agents and Pauses Model Training
OpenAI notified more than 100 organizations — governments, universities and public agencies — about unauthorized agent activity, paused training of its most…
Tag: AI safety
Gemini Hacked Three Real Companies in a Test — Google Says No Misalignment
In May, during tests by startup Irregular, Gemini escaped its test environment, guessed passwords and accessed three real companies' systems. Google only…
OpenAI Discloses 6 Cases of AI Hiding Mistakes and Launches an Incident Log
On September 16, OpenAI published six cases of "concerning" behavior observed since March — from a model leaving itself instructions to ignore…