Anthropic has announced the immediate suspension of access to its most advanced artificial intelligence models, Fable 5 and Mythos 5, after receiving a directive from the United States government to limit foreign access to the technologies due to national security concerns. The company confirmed on Friday that it had been instructed under export control measures to suspend availability of the models for foreign nationals. However, Anthropic stated that the government did not provide detailed evidence explaining the exact national security risk behind the decision. According to the company, officials believe there may be a method to bypass or “jailbreak” protections built into Fable 5, potentially allowing the model to assist in identifying software vulnerabilities. In response to the order and to ensure compliance, Anthropic said it had no option but to disable both models globally for all users, while clarifying that access to its remaining AI systems would continue without disruption.
The latest move arrives at a sensitive moment for Anthropic, which recently confidentially filed for a US initial public offering, placing it among leading AI firms preparing for public markets. Tensions between Anthropic and the US government have also been visible this year after the company reportedly declined requests to allow military use of its models for domestic surveillance and fully autonomous weapons systems. That disagreement resulted in Anthropic being added to a supply chain blacklist scheduled to take effect later this year. Friday’s directive signals a stronger shift in how Washington is approaching artificial intelligence oversight. Historically, US export controls largely focused on semiconductors and advanced computing hardware powering AI systems rather than restricting access to the software itself. Anthropic argued that the government relied only on verbal evidence of what it described as a “potential narrow, non universal jailbreak” and maintained that a limited vulnerability should not justify removing access to a commercial AI system already used by hundreds of millions of people.
The development also highlights growing debate between regulators and technology firms over the safety standards required for advanced AI deployment. Only days earlier, Anthropic publicly supported stronger government oversight, including measures to stop AI models considered excessively risky. Despite that stance, the company said the current directive does not align with fair or evidence based regulation. Pentagon Chief Information Officer Kirsten Davies publicly backed the move in a post on X, stating that national security priorities outweigh business interests and financial considerations. Dean Ball, a former White House official who contributed to the administration’s AI Action Plan in 2025, suggested the order could mean that all non Americans, including foreign nationals residing in the United States, may eventually need to verify citizenship to gain access to Anthropic’s newest models. Anthropic has not commented on whether foreign born staff members or researchers could face restrictions under the policy.
Anthropic launched Claude Fable 5 earlier this week as part of a new “Mythos class” category of advanced AI capabilities. The system included strict safety controls designed to limit use in high risk sectors such as cybersecurity, though some users criticised the restrictions as overly broad. Experts have warned that advanced models in the Mythos category could significantly increase the sophistication of cyberattacks, particularly in industries such as banking where legacy technology systems remain widely used. Anthropic said it coordinated with government agencies ahead of the Fable 5 launch and argued that rival AI systems also possess the ability to detect minor software weaknesses. Amazon Web Services later confirmed that Anthropic requested revocation of access to Fable 5 and Mythos 5 for users across all regions. A US Commerce Department official separately confirmed the directive, while Anthropic said it continues discussions with authorities in hopes of restoring access as quickly as possible, adding that applying similar restrictions industry wide could delay future frontier AI model rollouts.
Follow the SPIN IDG WhatsApp Channel for updates across the Smart Pakistan Insights Network covering all of Pakistan’s technology ecosystem.
