CISA is reportedly using Anthropic’s Mythos AI model to scan federal code repositories for security vulnerabilities, with the goal of finding and fixing flaws before adversaries can exploit them. The effort has allegedly already uncovered many bugs, while Anthropic’s broader relationship with US agencies has also faced tension over model safeguards and access restrictions. #CISA #Anthropic #Mythos #Pentagon #NSA #Fable
Keypoints
- CISA is using Anthropic’s Mythos AI model to audit federal software.
- The scans aim to find vulnerabilities before foreign intelligence agencies or cybercriminals can exploit them.
- CISA’s Attack Surface Evaluation team is leading the initiative.
- The AI-driven effort has reportedly uncovered a large number of software flaws.
- Anthropic has faced separate tensions with US officials over model safeguards and access controls.