The U.S. Department of Commerce notified Anthropic that export controls on its Fable 5 and Mythos 5 artificial intelligence models have been lifted, the company said in a statement Wednesday. The move reverses a June 12 order that forced Anthropic to suspend access to the models worldwide after the government raised concerns about potential misuse.
As MSI previously reported, Commerce Secretary Howard Lutnick announced a deal to restore Fable 5 access on June 30, after which the Commerce Department lifted the full restrictions. …
According to a letter from Lutnick to Anthropic seen by the BBC, the company “has agreed to proactively detect and address security risks associated with the models.” Anthropic also committed to collaborating on future releases of its AI models and to alerting the government of any malicious activity, according to the letter. The Commerce Department, Lutnick wrote, reserves the right to reconsider its decision and reimpose restrictions if necessary.
The BBC reported that it has contacted the Department of Commerce for further information. The letter did not detail what specific risks Anthropic has since addressed that satisfied the government’s concerns.
Fable 5 is a version of Anthropic’s Claude platform designed for the consumer market, capable of deep reasoning and performing complex tasks independently. Mythos 5 is a version built for businesses and cybersecurity experts.
The June 12 order came after U.S. authorities identified a method of bypassing Fable 5’s safety guardrails, according to Anthropic’s statements at the time. The company said the government had identified the method but had not demonstrated specific national security concerns that warranted the broad export restrictions.