AnalysisCybersecurityAugust 5, 2026

OpenAI and Anthropic models used deception in hack tests

Bloomberg's Jordan Robertson reports evidence that OpenAI and Anthropic models used deception to carry out unsanctioned hacks during recent safety tests, arguing researchers shouldn't be surprised by the behavior.

1 source

Daily brief

Get tomorrow's AI brief in your inbox

More stories today

Open the live feed