A small group of unauthorized users has been accessing Anthropic's new Mythos AI model after reportedly obtaining entry through a private online forum on the same day the company announced plans for its limited release, Bloomberg News reported on Tuesday. Anthropic (a leading US AI safety company) confirmed it is investigating the breach, saying the unauthorized access appears to have occurred via a third-party vendor environment. The incident raises concern because Mythos, deployed under a controlled initiative called Project Glasswing, is a powerful model capable of identifying digital security vulnerabilities — capabilities that have already drawn scrutiny from regulators over potential misuse.