Anthropic Restores Advanced AI Models After US Lifts Export Ban

The U.S. Department of Commerce has lifted export controls on Anthropic’s Claude Fable 5 and Mythos 5, ending a six‑day pause that began on 12 June after the agency raised national‑security concerns.

Claude Fable 5 is a consumer version of the firm’s flagship model, capable of deep reasoning and self‑directed task completion. Mythos 5, by contrast, is the enterprise‑grade variant used by cyber‑security experts to locate and exploit software vulnerabilities.

Anthropic said in a social‑media post that access will be restored to both models on Wednesday, following a notification that the export ban had been revoked.

The brief embargo stemmed from an alleged ability to “jailbreak” safeguards in Fable 5—allowing malicious users to bypass safety restrictions. The company argued that the reported loophole was narrow and that the coverage of the entire platform was unwarranted.

Federal officials had not identified a specific threat, yet the agency’s precautionary stance prompted the worldwide suspension of the two models.

With the controls lifted, Anthropic will resume commercial operations, but the incident highlights growing scrutiny over how advanced AI tools are regulated in the U.S.

Industry observers expect that the company will work with the Commerce Department to strengthen policy frameworks and clarify compliance for future launches.

Bloomberg reported that the decision signals increased confidence in throttling mechanisms for large language models, though experts caution that the border between safe deployment and potential abuse remains thin and contested.

The U.S. now faces the challenge of balancing innovation with protection against misuse as it revokes controls that were applied after a narrow security gap was identified.

For now, users of Claude and Mythos can expect full functionality, but the episode serves as a reminder that the regulatory landscape for AI remains fluid and dynamic.