Anthropic Says US Lifts Export Ban on Its Advanced AI Models

Anthropic says the United States has lifted the export ban on its advanced AI models, BBC News reported on July 1, after the company's Fable and Mythos models were abruptly suspended from export in June over concerns they could be used by hackers.

CNBC's coverage states that the Trump administration lifted export controls on Claude Fable 5 and Mythos 5, and The Wall Street Journal reported on July 1 that Anthropic had neared a deal with the administration to restore access to the Fable model. The Guardian also covered the lifting of the controls the same day, making this one of the most heavily corroborated AI-governance stories of the cycle.

The stated rationale for the June suspension — that the models could be used by hackers — is attributed to the BBC's account; the material reviewed did not include the government's formal notice, the terms under which access was restored, or any conditions Anthropic accepted as part of the arrangement the Journal described.

Background

Anthropic, founded in 2021 by former OpenAI researchers including siblings Dario and Daniela Amodei, is one of a small group of frontier AI laboratories, and its Claude family of models is deployed widely across enterprises and government. The company has cultivated a reputation for emphasizing safety research, which makes a security-grounded export restriction on its own models a notable inversion of its usual positioning in policy debates.

U.S. export controls have historically bitten hardest on the hardware layer of the AI stack — advanced chips and chipmaking tools, administered by the Commerce Department's Bureau of Industry and Security. Extending controls to the models themselves, and then unwinding them weeks later through direct negotiation with the developer, would mark a different kind of instrument: one aimed at software capabilities rather than silicon, and applied to an American company's flagship products.

What comes next

The terms of the restoration are the open question. Watch for any formal notice from the administration setting out conditions on the models' export, for Anthropic's own account of what it agreed to, and for whether the June suspension-and-release pattern becomes a template applied to other frontier developers — none of which is established in the reporting to date.