Commerce Dept. Lifts Export Controls on Anthropic AI Models

The Commerce Department is set to lift export controls on Anthropic's two most advanced AI models following negotiations between the company and the Trump administration, according to a person familiar with the matter.

The restrictions on both the Fable 5 model and the more capable Mythos 5 model will be removed, ending a dispute that had limited the company's ability to distribute its technology. Mythos 5 had previously been approved only for select companies and government agencies.

The deal

Commerce Secretary Howard Lutnick is expected to communicate the decision in a letter to Anthropic cofounder Tom Brown. Lutnick has worked alongside national cyber director Sean Cairncross to resolve the standoff with the AI company.

The breakthrough came after Anthropic shifted its approach to addressing the administration's concerns about users bypassing safety restrictions—particularly those related to cybersecurity capabilities. The company initially argued that the government's worries about jailbreaks were excessive and that preventing all workarounds was technically impossible.

In recent weeks, however, Anthropic changed strategy. Rather than debating whether jailbreaks could be completely eliminated, the company committed to building more robust safeguards to reduce their frequency. This pragmatic shift aligned with what administration officials wanted to hear.

Why it matters

The resolution demonstrates how AI companies must balance technical reality with regulatory expectations. Export controls on AI models represent a new frontier in technology policy, where governments are attempting to restrict the flow of advanced capabilities that could pose national security risks. Anthropic's experience shows that successful navigation of these controls may depend as much on communication strategy and relationship management as on technical measures. The case also highlights the growing role of jailbreak prevention—the ability to stop users from circumventing safety guardrails—as a key regulatory concern for frontier AI systems.

Leadership changes

The negotiations also involved personnel adjustments. Anthropic CEO Dario Amodei was replaced in meetings by cofounder Tom Brown, whom administration officials reportedly preferred on a personal level, according to WIRED, which first reported these details.

The dispute had created uncertainty for Anthropic as it sought to compete in the global AI market while navigating new regulatory frameworks designed to control the export of powerful AI capabilities. The company's willingness to adapt its communication approach and commit to enhanced safety measures ultimately cleared the path for the restrictions to be lifted.

These details were first reported by WIRED.