Anthropic said on June 30 that the US government had lifted export controls on its Fable 5 and Mythos 5 models, clearing the way for the company to start restoring access after less than three weeks of restrictions. Fable 5 would return globally starting July 1, while Mythos 5 was already restored for a set of approved US organizations.
The export controls were originally imposed on June 12, when the US government ordered it to suspend access to Fable 5 and Mythos 5 for foreign nationals. Anthropic had no reliable way to verify nationality in real time, so it suspended access to both models for all users.
On June 26, the US government allowed Mythos 5 to return for a limited set of trusted US organizations. Now, this access had already been restored for that group, and it was continuing to work with the government to widen access through the Glasswing program.
[ALSO READ: Anthropic Restarts Limited Mythos 5 Rollout After US Review ]
According to Anthropic, the US government’s concerns were tied to a reported jailbreak technique that could bypass Fable 5’s safeguards and prompt the model to identify software vulnerabilities. The company worked with the government and with Amazon to review the report, then built a new safety classifier to block the technique. Anthropic said the new classifier blocks the reported method in more than 99% of cases.
The specific technique described in the Amazon report did not expose unique Mythos-level cyber capabilities. The behavior was closer to a borderline case for Fable 5’s safeguards, and the same vulnerabilities could also be identified by less capable models in testing.
Who gets access now
Anthropic said Fable 5 will be available starting July 1 to users globally on the Claude Platform, Claude.ai, Claude Code and Claude Cowork. For Pro, Max, Team and select Enterprise plans, it will be included for up to 50% of weekly usage limits through July 7, after which it will be available through usage credits.
Anthropic said they will re-enable access on AWS, Google Cloud and Microsoft Foundry as quickly as possible.
Anthropic has restored access for a set of US organizations after government approval.
Why the US intervened
According to Reuters, the US was concerned that advanced AI models could be misused by foreign military intelligence or other actors in countries of concern. The models were treated differently because Mythos 5 is the more permissive version for defensive cybersecurity use, while Fable 5 is the broadly available public model with stronger safeguards.
The government reserved the right to restore controls if situations change or Anthropic fails to meet its commitments. Those commitments include proactively detecting and handling security risks, notifying the US about malicious activity and working with the government on safeguards for current and future models.
[ALSO READ: Before the Public Sees Them, the U.S. Government Will Test Top AI Models ]
Now, the industry is moving toward formal security review for powerful AI models. Anthropic is working with Amazon, Microsoft, Google and other partners on a shared framework for judging jailbreak severity, and it has proposed more direct government collaboration on pre-release testing and information sharing.
A timeline
- June 9: Fable 5 and Mythos 5 launched.
- June 12: US export controls forced a full suspension.
- June 26: Mythos 5 was approved for a limited set of US organizations.
- June 30: Commerce lifted export controls.
- July 1: Fable 5 returns globally, with staged access on some paid plans.
- Anthropic said the new safeguard blocks the reported jailbreak technique in over 99% of cases.
