Anthropic PBC has halted access to its most advanced artificial intelligence models following a directive from the U.S. government restricting use by foreign nationals. The Commerce Department, under the Trump administration, ordered Anthropic to suspend access to its Fable 5 and Mythos 5 AI systems for all non-U.S. individuals, regardless of location, citing national security concerns. The company subsequently closed off access to the affected models for all customers to ensure compliance, while maintaining availability of its popular Claude AI.

This government intervention marks an unprecedented move to broadly restrict foreign access to cutting-edge AI software developed by an American firm. While previous U.S. administrations have imposed export controls on hardware technologies such as semiconductors and supercomputers, applying similar restrictions directly to AI software is a novel development that has sparked debate around constitutional, commercial, and strategic implications.

Anthropic stated it disagrees with the government’s decision, arguing that the restriction stems from a narrow “jailbreak” vulnerability in Fable 5—a technique potentially bypassing built-in safety guardrails. The company contended that if this rationale were applied industry-wide, it could effectively freeze progress in deploying frontier AI models. Nonetheless, Anthropic is complying with the directive amid growing tensions between the firm and the Trump administration, stemming from earlier disputes over military and surveillance applications of Anthropic’s technology. The administration previously designated the company a supply-chain risk and ordered agencies to phase out its products.

The restrictions also reverberate internationally, prompting concerns from the European Union about technological sovereignty and cybersecurity risks posed by advanced AI systems. Several European officials have emphasized the need for increased autonomy in technology development as new AI capabilities evolve rapidly.

In the United Kingdom, political figures voiced concerns over the export control’s impact on the nation’s AI ambitions. Former Conservative security minister Tom Tugendhat described the ban as highlighting the increasing importance of digital sovereignty, where control over AI technology is now a strategic factor in national power. Shadow business secretary Andrew Griffith called the move “a disaster for Britain,” criticizing the government’s inability to secure exemptions and warning that it hampers the UK’s efforts to become a global AI leader. The ban has disrupted ongoing work by British researchers, companies, and healthcare institutions that had been piloting Anthropic’s models.

The UK government has launched a sovereign tech fund aimed at bolstering domestic AI capabilities, but critics argue that its £500 million budget falls short of the substantial investment necessary to compete globally. Labour’s tech minister Kanishka Narayan highlighted the country’s strong academic foundations and growing AI industry presence, citing initiatives such as Google DeepMind and new Anthropic offices in London as evidence of the sector’s growth.

Anthropic’s CEO Dario Amodei had previously advocated for governmental authority to restrict access to AI models deemed to pose extreme risks but emphasized the need for balanced approaches to mitigate harm without stifling innovation. Industry leaders, including Nvidia CEO Jensen Huang and OpenAI CEO Sam Altman, have generally encouraged the U.S. government to foster widespread adoption of American AI systems internationally to maintain technological leadership.

The export control order intensifies ongoing debates about how best to regulate AI development and deployment amid rapid advancements. It also underscores geopolitical tensions as countries seek to safeguard sensitive technology while promoting innovation in a field with far-reaching economic and security implications.