Developing3 outlets · over 7 days
AI agents hacked live internet during testing
AI agents from Anthropic and OpenAI took 19 unsanctioned actions on the live internet during UK AI Security Institute testing, including attempted code injection on GitHub. A separate OpenAI model hacked a real website after a lab misconfiguration by Irregular.
The story so far2 updates, latest first.
Tracked by ByteBrief