Anthropic Disables Advanced AI Models Following US National Security Directive

Anthropic, an artificial intelligence developer, has disabled its most advanced AI models, Fable 5 and Mythos 5, for all users globally. This decision comes after the US government issued an export control directive requiring the company to suspend access to these models for foreign nationals, citing national security concerns.

According to Anthropic's statement, the government did not provide specific details regarding its national security concerns. However, Anthropic understands that the government believes a method, referred to as a "jailbreak," could bypass a safeguard designed to prevent Fable 5 from being used to identify software vulnerabilities.

Anthropic has expressed its disagreement with the directive, stating that the finding of a "narrow potential jailbreak" should not warrant recalling a commercial model deployed to a large user base. The company believes there is a "misunderstanding" and is working to restore access to the models as soon as possible. Anthropic also stated that applying such a standard across the industry would "essentially halt all new model deployments for all frontier model providers."

This government action marks an escalation in US efforts to control foreign adversaries' AI capabilities, shifting focus from merely restricting access to AI-powering chips and tools to limiting access to the AI models themselves. The order follows a period of tension between Anthropic and the US government, which previously saw Anthropic placed on a supply chain blacklist after refusing to allow its AI models to be used by the US military for domestic surveillance or fully autonomous weapons systems.

The Claude Fable 5 model, a "Mythos-class" AI, was rolled out earlier in the week with guardrails to prevent its use in risky areas such as cybersecurity. However, experts have cautioned that such Mythos models, if misused, could significantly accelerate sophisticated cyberattacks, particularly against complex and often legacy technology systems found in sectors like banking. Anthropic had engaged with the US government and others on safety prior to the Fable launch.

A US official confirmed that the Commerce Department had issued the export control directive. Kirsten Davies, the Pentagon’s chief information officer, stated on X that the Defense Department supports prioritizing national security. Dean Ball, a former White House official, suggested on X that the order could restrict all "non-Americans," including those based in the US, from using Anthropic’s latest models, implying a need for citizenship verification.

Anthropic confidentially filed for a US initial public offering (IPO) last month.

According to Dawn Pakistan, Amazon's cloud unit AWS confirmed that Anthropic requested it revoke access to the models for "all users in all regions."