Anthropic restores Fable 5 access after export controls ease
The AI company says U.S. export limits on Fable 5 have been relaxed after talks with the Trump administration over security risks.
By Maya Lindqvist · Senior Technology Correspondent
3 min read
Anthropic has brought Fable 5 back online globally after the Trump administration eased export controls that had restricted access to the company’s most powerful AI model. The move matters because the two-week shutdown showed how quickly Washington can disrupt the release of frontier AI systems on national security grounds.
Anthropic said Tuesday that the controls had been relaxed. Commerce Secretary Howard Lutnick also confirmed the shift on X, saying the government had worked with Anthropic over the past two weeks to ensure Fable was aligned with U.S. interests.
Access to Fable and a related Anthropic model, Mythos 5, had been restricted since mid-June, Fortune reported. The U.S. government first barred use by foreign nationals, a rule that applied regardless of whether users were inside or outside the United States and would have covered foreign Anthropic employees.
Anthropic then suspended access for all users, according to Fortune. The Trump administration said it acted because of national security concerns, including the risk that hostile actors could break through model safeguards meant to block malicious use.
The restrictions began to loosen last week. Semafor reported that the government allowed Anthropic to make Mythos available to a selected group of more than 100 U.S.-based companies and federal agencies.
Fortune reported that the staged release resembled Anthropic’s own earlier rollout plan. The company first gave Mythos to a limited set of users, then released Fable as a model with the same underlying capabilities but more guardrails and safety filters for broader use.
With Fable and Mythos available again to their intended users, Anthropic is returning to roughly the same product setup it had before the export controls took effect, according to Fortune. The company’s public statement also pointed to closer cooperation with the White House after the dispute.
Anthropic said it would “scale up our government collaboration” and work with the administration and private-sector companies on a “shared industry framework.” The company named Amazon, Microsoft and Google as partners in an effort to set standards for testing vulnerabilities in frontier models.
Under that plan, Anthropic said security researchers and industry partners would use a common system for rating the severity of jailbreaks and a shared playbook for responding to them. The company also said it would give federal authorities early access to frontier models so they can review them and test for weaknesses before release.
Anthropic said it would create dedicated research teams to help ensure its systems do not conflict with government priorities. The company said it hopes the effort can form the basis for industrywide rules and a broader model for coordinating AI risks and benefits.
The reset follows months of tension between Anthropic and federal officials. Fortune reported that the Defense Department labeled Anthropic a supply chain risk earlier this year after the company refused to let the Pentagon use its models for surveillance or autonomous warfare operations.
Politico reported that Anthropic CEO Dario Amodei resisted several Trump administration requests, including a request that the company voluntarily take Fable off the market. Anthropic, which said in May it was valued at $965 billion in its latest funding round, has also filed confidential paperwork with the SEC for an initial public offering, Fortune reported.
This story draws on original reporting from Fortune.