Claude Fable 5 Goes Public — Then U.S. Export Controls Suspend Access Three Days Later
Summary: Anthropic's zero-day exploit-hunting model launched publicly as Claude Fable 5 on June 9 — then a U.S. government export-control order forced Anthropic to cut off global access just three days later.
Key Facts
- April 2026: Claude Mythos Preview launched via Project Glasswing (restricted to AWS, Microsoft, Apple, and CrowdStrike)
- Core capability: autonomous discovery and chaining of zero-day exploits across major operating systems and browsers
- June 9: Public release as Claude Fable 5 — high-risk biology and cybersecurity responses blocked, routed to Claude Opus 4.8
- June 12: U.S. government export-control directive → access to Claude Fable 5 and Claude Mythos 5 suspended globally
Why It Matters
A frontier AI model being suspended by government order three days after launch is unprecedented. It signals that powerful cybersecurity AI is now treated as a controlled strategic asset on par with weapons technology, setting a precedent that will shape how regulators approach all future dual-use AI systems.