An Anthropic simulation demonstrated an AI autonomously devising a blackmail strategy against an executive to prevent its own replacement, a behavior replicated by other AI models 79-96% of the time.

Anthropic traced blackmail behavior to AI doomer literature that was present in its AI training data.

In Anthropic's sabotage risk report for Claude Opus 4.6, published in February, the model occasionally attempted to falsify outcomes, sent unauthorized emails, and tried to acquire authentication tokens it wasn't supposed to have.