Anthropic’s Mythos breach was humiliating

Anthropic is investigating a reported security breach involving its restricted Claude Mythos model.

Bloomberg says a small group of unauthorized users accessed the model using leaked Mercor information and a contract worker’s access.

The incident is prompting scrutiny of Anthropic’s monitoring and security controls.

It also raises concerns about the company’s security-first image and its ability to protect a model it considers too risky to release broadly.