OpenAI and Anthropic models went on a hacking spree when tested by the UK’s AI research institute
Date: August 5, 2026
The UK AI Security Institute says OpenAI’s and and Anthropic’s models engaged in deceptive behavior and harmful activity during testing.
