Trump Administration Blocks Foreign Access to Anthropic AI Models

The Trump administration has ordered Anthropic to suspend foreign access to its newest AI models, Fable 5 and Mythos 5, citing national security concerns. The directive requires the San Francisco-based company to block all foreign nationals from using the models, whether they are located inside or outside the United States.

According to Anthropic's Friday blog post, the company received an export control directive from the US government requiring immediate compliance. The order came less than a week after Anthropic released the new models based on its Mythos technology platform.

Why it matters

This marks an unprecedented expansion of US export controls into software access rather than just hardware sales. The directive could reshape how American AI companies operate globally, potentially forcing them to verify user nationality and creating compliance burdens that may advantage competitors in other countries. For businesses relying on cutting-edge AI tools, the order introduces new uncertainty about access to advanced models.

Security Concerns Behind the Order

The directive stems partly from suspicions that a China-linked group accessed Anthropic's new AI model, according to reporting by Semafor. David Sacks, an adviser to President Trump, stated on X that the government received warnings that Fable 5 could be "jailbroken" — a term referring to bypassing safety restrictions built into AI systems.

Sacks claimed that when Anthropic CEO Dario Amodei was notified about the vulnerability, the company failed to address it. Anthropic disputed this characterization, stating the government provided only "verbal evidence of a potential narrow, non-universal jailbreak" and arguing that such a limited finding should not warrant recalling a model deployed to hundreds of millions of users.

Experts told Reuters that Mythos models could accelerate sophisticated cyberattacks if misused, particularly against banking systems that rely on complex, interconnected legacy technology.

Broader Implications for Research and Business

The order extends beyond consumer access. Foreign nationals working in the US on H1-B visas will lose access to the models, as will international research institutions collaborating with American partners. Several key Anthropic personnel, including co-founder Chris Olah and AI researcher Andrej Karpathy, were born outside the US, though it remains unclear whether they face restrictions.

Companies integrating Anthropic's Claude software into their operations will face disruption. S&P, for example, uses Claude to help financial advisers and analysts access data more efficiently. US companies with international workforces may find some employees unable to use the advanced models, creating productivity challenges.

The directive represents an escalation in the ongoing dispute between Anthropic and the Trump administration. The company is currently suing the administration after being placed on a supply chain blacklist for refusing to allow the US military to use its AI models for domestic surveillance and fully autonomous weapons systems.

Anthropic stated it believes there is a "misunderstanding" and is working to restore access to the models. Access to the company's other AI models remains unaffected by the order.

These details were first reported by Al Jazeera, with additional reporting from Semafor.