A sweeping U.S. government directive has forced Anthropic to suspend customer access to its most advanced artificial intelligence systems, marking a watershed moment in the growing clash between national security priorities and frontier AI development. The company announced Friday that access to its Mythos 5 and Fable 5 models has been disabled after federal authorities ordered restrictions tied to security concerns.
Government Order Triggers Broad Suspension
Anthropic said it received a directive from the U.S. government requiring the suspension of access to the models by foreign nationals. Instead of attempting a limited rollout or selective restrictions, the company removed customer access altogether while complying with the order.
The move ranks among the most aggressive government interventions yet involving a commercially deployed AI model. Anthropic said officials cited national security concerns but did not provide detailed public findings supporting the decision.
The Commerce Department, which issued the restriction, did not immediately comment. Axios reported that Anthropic would be required to obtain licenses for the export, re-export, or domestic transfer of the affected models.
Jailbreak Concerns Draw Government Scrutiny
Anthropic believes the government’s concerns may be linked to a technique capable of “jailbreaking” Fable 5, allowing users to bypass some of the model’s built-in safeguards.
According to the company, demonstrations of the method revealed only a limited number of previously identified vulnerabilities. Anthropic described the weaknesses as minor and argued that similar issues can be discovered by other publicly available AI systems without requiring a safety bypass.
The company also emphasized that it introduced multiple layers of protection aimed at reducing cybersecurity-related misuse and had previously worked alongside government teams to stress-test the model’s defenses. Anthropic maintains that no advanced AI system is completely immune to jailbreak attempts.
Anthropic Pushes Back Against the Decision
While complying with the directive, Anthropic strongly disputed the reasoning behind it. The company argued that a narrowly defined jailbreak pathway should not justify withdrawing access to a model deployed at massive scale.
Anthropic warned that applying the same threshold across the industry could significantly slow future AI releases. The restriction also impacts many foreign nationals working within the company, limiting their ability to interact directly with the affected systems.
As AI policy analyst Daniel Castro of the Information Technology and Innovation Foundation has argued in broader debates around AI governance, regulators face the challenge of balancing legitimate security concerns without creating barriers that stifle innovation.
Longstanding Tensions With the Trump Administration
The latest dispute adds another chapter to Anthropic’s complicated relationship with the Trump administration. Earlier this year, the company was designated a “supply chain risk” in military procurement discussions tied to AI safety requirements for defense applications.
Anthropic challenged the designation in court and secured an early legal victory. Despite those disagreements, federal officials remained engaged with the company, particularly after Mythos attracted attention for cybersecurity capabilities that unsettled both policymakers and investors.
The scrutiny reflects a wider trend. According to the OECD AI Policy Observatory, governments worldwide have introduced more than 1,000 AI-related policy initiatives and regulatory measures in recent years, underscoring how rapidly oversight of advanced AI systems is expanding.
A Defining Moment for the AI Industry
The controversy arrives as governments take a more active role in monitoring increasingly capable AI models. Mythos also helped shape a recent executive order encouraging companies to share highly advanced cyber-focused systems with federal authorities before broader deployment.
For the AI industry, the suspension highlights how national security considerations are becoming a central factor in decisions about model access, deployment, and governance. As GrowBusinessMag and other industry observers watch developments unfold, the outcome could establish a powerful precedent for how governments respond to future breakthroughs. The decisions made in the coming months may influence not only Anthropic’s roadmap but also the regulatory framework governing frontier AI across the global technology sector.