Redeploying Fable 5

Details

On June 12, the US government applied export controls to Claude Fable 5 and Mythos 5 without advance notice, forcing Anthropic to suspend access for all users globally — the company had no real-time nationality verification mechanism. The controls were lifted on June 30; Fable 5 returned July 1 for all global users, while Mythos 5 remains restricted to a set of approved US organizations after government sign-off on June 26. Anthropic used the two-week shutdown to update Fable 5's cybersecurity safeguards, deploying new safety classifiers designed to block dangerous cyber uses while preserving legitimate defensive security work. The company also initiated a HackerOne program for security researchers and developed a shared jailbreak severity framework with Amazon, Microsoft, Google, and other Glasswing partners. The redeployment plan covers Fable 5 in up to 50% of weekly usage limits for Pro, Max, Team, and select Enterprise plans through July 7, after which it transitions to usage credits. AWS, Google Cloud, and Microsoft Foundry integrations were to be re-enabled as quickly as possible. This episode is the most significant forced model suspension by a US AI lab to date. Anthropic announced deeper government collaboration going forward: pre-release testing with US agencies, increased information sharing, and joint research. The episode is being treated as a template for how AI labs and the US government should co-develop AI governance procedures.