Anthropic Clashes With U.S. Over AI Export Controls on Mythos

The Trump administration forced Anthropic to recall its latest AI models this month after imposing unprecedented export controls, marking a significant escalation in the government's willingness to intervene in frontier AI development.

On June 12, the administration banned foreign nationals from accessing Anthropic's Mythos 5 and Claude Fable 5 models. Anthropic had released Mythos only to vetted organizations, while Fable was a public version with additional safeguards. To comply with the order, the company disabled access to both systems entirely.

According to CNBC, Amazon alerted the administration to security concerns with Fable shortly after its release. The Trump administration, already worried about Mythos's cyber capabilities, quickly directed Anthropic to suspend access. Senior Anthropic representatives met with administration officials in Washington on Monday, but the restrictions remain in place.

The security concern

Anthropics stated publicly that the government hasn't specified its exact concern, though the company believes it relates to potential workarounds of the model's cybersecurity safeguards. Anthropic maintains this vulnerability is minor and already exists in other publicly available models.

The dispute centers on whether the models could enable foreign countries to conduct large-scale cyberattacks against the United States. After Anthropic's partial release of Mythos, President Trump signed an executive order requiring major AI companies to voluntarily submit cutting-edge models for a 30-day government review period.

A new regulatory dynamic

"AI is a general purpose technology that is reshaping the world," Michael Horowitz, senior fellow for technology and innovation at the Council on Foreign Relations, told the Monitor. "The frontier AI labs in the United States have enormous power and influence, but so does the U.S. government."

The speed of government action represents a departure from typical defense bureaucracy. "Usually the problem that we have with government action [on defense] is that it is way too slow," said Emily Kilcrease, director of the Energy, Economics, and Security program at the Center for a New American Security. "What happened over this past week with Anthropic was interesting to see that the government moved very quickly."

Why it matters

This confrontation demonstrates that national security concerns have fundamentally shifted the AI regulation debate. The government now wields significant leverage over AI companies, and the question is no longer whether to regulate but how much. Unlike the hands-off approach taken during the internet's development, AI's military and cybersecurity implications mean companies cannot expect minimal oversight—even as they possess expertise the government desperately needs.

Companies hold unique leverage

Despite tensions, both sides continue collaborating. The government met with Anthropic representatives to discuss Mythos's capabilities after its March release, and Anthropic worked with officials to test safeguards before releasing Fable 5.

Companies like Anthropic maintain leverage because few people understand how frontier models actually work. Kilcrease estimates only about 200 engineers worldwide possess sophisticated knowledge of these AI systems—and they don't work for the government. "[The developers] are always going to be the ones that have the most recent, up-to-date, in-depth knowledge about how the technology works," she said.

The relationship remains complicated. In March, the Pentagon declared Anthropic a supply chain risk when the company refused to give the government unlimited access to its models for defense purposes.

These details were first reported by The Christian Science Monitor.