Anthropic can now share its most advanced AI model with over 100 companies after the U.S. government lifted its export ban on Claude Mythos 5, a spokesperson for Anthropic confirmed in a statement to The Hill.
The decision marks a major reversal from June 12, when the federal government imposed restrictions on Mythos 5 and its weaker counterpart, Fable 5, citing security concerns with these models. Commerce Secretary Howard Lutnick notified the company of the change in a letter obtained by The Hill.
"Since the issuance of my June 12 letter, Anthropic has worked with the U.S. government to address risks associated with Covered Models," Lutnick wrote, referring to the Mythos and Fable models. "These efforts have yielded significant progress."
What the Right Is Saying
Conservative observers have praised the administration's approach as a pragmatic balance between security concerns and economic competitiveness. Supporters argue that restricting American AI capabilities too aggressively risks ceding market share to foreign competitors who face fewer regulatory constraints.
Business groups contend that managed access to advanced models through vetted partners represents an appropriate middle ground. They note that Anthropic's willingness to work with federal regulators on "protocols and standards" demonstrates how public-private cooperation can address security concerns without sacrificing innovation or international competitiveness.
What the Left Is Saying
Progressive policy advocates argue that while the government's reversal reflects constructive engagement with Anthropic, the episode underscores the need for permanent guardrails on advanced AI exports. Consumer protection groups have long maintained that releasing powerful dual-use AI systems to dozens of partners creates compounding security vulnerabilities that individual company commitments cannot fully address.
Civil liberties organizations contend that export control frameworks should require ongoing independent audits rather than relying on self-reported compliance from private firms. They note that the list of approved partners remains subject to unilateral revision by the Commerce Secretary, raising questions about transparency and accountability in how access decisions are made.
What the Numbers Show
The original June 12 restrictions affected two Anthropic models: Mythos 5, the company's most advanced system, and Fable 5, a less powerful counterpart. Friday's action restores access to Mythos 5 for more than 100 approved companies. The restrictions on Fable 5 remain in place, according to Lutnick's letter.
Anthropic stated it is "working to provision the approved set of providers and restore their access to Mythos 5 as quickly as possible." The company added that it continues to work with the government toward expanding access and making Fable 5 available for general use again.
Lutnick emphasized in his letter that he can amend the list of approved partners "at any time" and reserves the right to reevaluate licensing requirements "should circumstances change."
The Bottom Line
The decision arrives on the same day rival AI company OpenAI released its own powerful model, GPT-5.6, to government-approved partners, highlighting intense competition in the advanced AI sector.
What happens next: Watch for how quickly Anthropic can restore access for approved partners and whether Fable 5 restrictions are lifted in coming weeks. The scope of the partner list and any future amendments will signal how broadly the administration interprets "trusted partners."
Anthropic has committed to ongoing collaboration with federal officials on protocols and standards, suggesting this episode may set precedents for how export controls handle advanced AI systems going forward.