Anthropic to Shut Down Advanced AI Models Following US Foreign Access Restrictions

Anthropic has confirmed that it will disable access to its latest AI models, Fable 5 and Mythos 5, after receiving a directive from US authorities restricting their availability to foreign nationals.

According to the company, the order was issued on national security grounds. Anthropic stated that officials did not provide detailed technical evidence explaining the specific risks associated with the models.

Concerns Surrounding AI Safety Protections

Anthropic said government officials expressed concerns that the safeguards embedded within Fable 5 could potentially be bypassed through a so-called AI jailbreak.

The company noted that the evidence presented involved what it described as a limited and highly specific scenario rather than a broadly exploitable vulnerability. Anthropic argues that such findings do not warrant removing a commercially deployed AI system from users.

Growing Friction Between Anthropic and Federal Authorities

The latest directive comes amid increasing tensions between Anthropic and policymakers in Washington.

Previous disagreements reportedly arose after the company resisted requests involving domestic surveillance applications and fully autonomous military systems. Those disputes were followed by reports that Anthropic could face restrictions related to future government procurement programs.

A New Phase in US AI Export Policy

The action signals a broader shift in how the United States approaches AI-related export controls.

Historically, restrictions focused primarily on advanced chips, computing infrastructure, and semiconductor technologies. The new directive directly targets access to advanced AI models themselves, expanding the scope of export-control policy.

Debate Over AI Jailbreak Assessments

The situation underscores ongoing disagreements between regulators and AI companies over how security vulnerabilities should be evaluated.

Anthropic has previously supported stronger oversight measures for advanced AI systems and has advocated for mechanisms capable of limiting access to high-risk models. However, the company maintains that the current order lacks sufficient evidence and transparency.

National Security Prioritized

Government officials defended the decision as a precautionary measure intended to safeguard national interests.

Pentagon Chief Information Officer Kirsten Davies stated that security considerations must remain the primary concern when evaluating access to highly capable AI technologies, even when such decisions affect business operations and commercial growth.

Consequences for Fable 5 and Mythos 5 Users

Fable 5 was recently introduced as part of Anthropic's premium Mythos-class family of AI systems. The models include safeguards designed to limit use in sensitive cybersecurity and security-related applications.

Cybersecurity specialists have repeatedly warned that increasingly capable AI systems could accelerate the discovery of software weaknesses and potentially aid sophisticated cyberattacks if deployed irresponsibly.

To comply with the government directive, Anthropic said it will disable access to both Fable 5 and Mythos 5. The company emphasized that its remaining AI products and services will continue operating normally.

What Happens Next?

Anthropic believes the directive may stem from misunderstandings regarding the actual risks posed by the models. Company representatives say discussions with regulators are ongoing and that they hope to restore access to Fable 5 and Mythos 5 as soon as possible.

A US official confirmed that the Commerce Department issued the export-control order that resulted in the suspension of foreign-national access to the two AI systems.