CISA reportedly using Anthropic's Mythos to scan government software for flaws

The US CISA is using Anthropic's Mythos AI model to audit federal code for vulnerabilities, led by its Attack Surface Evaluation team. The initiative has already uncovered a large number of flaws, though specific details remain undisclosed. Anthropic and CISA have not officially commented on the operation.