1 of 1
Story summary
- The NSA lost access to Anthropic’s Mythos AI model after June 2026 export controls.
- In Project Glasswing, Mythos flagged vulnerabilities in sensitive U.S. government systems.
- Senator Mark Warner said the model breached all classified systems within hours, though officials said it did not necessarily exploit the flaws.
- The export ban halted Mythos 5 and Fable 5 releases, placed Anthropic on a U.S. security blacklist, and sparked a debate on frontier AI in cybersecurity.
