Anthropic is preparing to restore access to its Mythos and Fable AI models after the U.S. Commerce Department removed export restrictions that had forced the company to pause availability of its most advanced systems earlier in June.
The reversal comes less than three weeks after Anthropic disabled access to Mythos 5 and Fable 5 following a U.S. export-control order tied to national security concerns.
The restrictions were part of Washington’s growing effort to monitor frontier AI models that could be misused by military, intelligence or cyber-linked actors in countries such as China and Russia.
According to Reuters, the Commerce Department withdrew the controls after Anthropic agreed to work more closely with U.S. officials on safety protocols, model-release standards and risk monitoring. Commerce Secretary Howard Lutnick said in a letter to the company that a license was no longer required to export Mythos or Fable.
Anthropic said it would begin restoring access after the decision, following an earlier partial easing in which U.S. officials allowed Mythos 5 to be made available to selected trusted American organizations before lifting the broader restrictions.
Why the models were restricted, and what changed
The original order reflected concern inside Washington that increasingly capable AI systems could create new security risks if released too widely without oversight.
Advanced models can support legitimate research and business use, but officials worry they may also help hostile users speed up cyber operations, analyze vulnerabilities or support sensitive technical work.
Mythos and Fable became part of that debate because they were considered among Anthropic’s most advanced releases, and their temporary suspension showed how quickly AI policy is moving from broad safety discussions into direct government intervention.
The latest decision does not mean the U.S. government has dropped its concerns. Instead, it points to a compromise: Anthropic can resume access, but under closer cooperation with federal officials. The company has agreed to detect and address security risks, coordinate on future release standards and report malicious activity linked to its models.
The dispute also shows the tension facing the AI industry, where stronger testing and safeguards may help reduce national security risks, while some technology leaders warn that government control over model access could shape competition, customer choice and the commercial rollout of frontier systems.
