Anthropic's most advanced AI models are back online worldwide after the U.S. government reversed the emergency export restrictions that had forced them offline. According to MeriTalk, the Commerce Department lifted the curbs on the company's Claude Fable 5 and Mythos 5 systems.
The restrictions traced back to safety concerns at the highest level of government. Ars Technica reports that the models had "spooked Trump into safety testing," and CBS News examined why Washington initially clamped down on the technology. The result, as several outlets describe it, was effectively a global shutdown of the models.
Anthropic moved quickly once the controls were lifted. MarkTechPost reports the company redeployed Claude Fable 5 on July 1 and added a new cybersecurity classifier as part of the relaunch. The company announced the return on its own social channels, and the news topped Hacker News under the headline "Fable 5 Is Back," drawing 141 points and dozens of comments. 9to5Mac confirmed the model is now available again.
The episode also touched on geopolitics. The Register reports that Anthropic is removing what it called "covert code for catching Chinese competitors," a detail tied to the broader story of how these frontier models were governed and secured.
Taken together, the sources describe a rare sequence: a leading AI developer pulling its top products off the global market under government pressure, submitting to safety testing, and then restoring access once regulators were satisfied.
Why it matters: The reversal shows that frontier AI is now powerful enough to trigger national-security-style export controls—and that the rules governing which models the world can use can change almost overnight.