U.S. Government Orders AI Firm Anthropic to Halt Access to Advanced Models Over Security Concerns
The U.S. government has issued an emergency directive ordering AI company Anthropic to immediately suspend access to its two most powerful artificial intelligence models, Claude Fable 5 and Claude Mythos 5, for all foreign nationals. The order, citing national security concerns, was issued on Friday, just days after the models were released.
The directive prohibits any foreign national, whether inside or outside the United States, from accessing the models. To ensure compliance with the sweeping order, Anthropic was compelled to disable the models for its entire customer base.
While the government’s directive did not specify the exact nature of the national security concern, it is understood that officials believe a method for bypassing or “jailbreaking” the publicly available Fable 5 model has been discovered. Mythos 5, which has fewer safeguards and is particularly adept at identifying cybersecurity vulnerabilities, was only accessible to a select group of partners.
Anthropic has disputed the severity of the government’s findings. The company stated that it reviewed a demonstration of the alleged bypass technique and concluded that the identified vulnerabilities appear relatively simple. Furthermore, Anthropic asserted that other publicly available AI models can achieve similar results without requiring any bypass.
The company also indicated that the government has so far provided only verbal evidence of a potential, narrowly defined, and non-universal jailbreak. This bypass reportedly involves instructing the model to read a specific codebase and identify software flaws. Anthropic claims to have verified that the capabilities demonstrated are already widely available from competing models, including OpenAI’s GPT-5.5.
While complying with the directive, Anthropic expressed concerns that the action sets a dangerous precedent. “If this standard was applied across the industry,” the company stated, “we believe it would essentially halt all new model deployments for all frontier model providers.” Access to all other Anthropic models remains unaffected, and the company is working to restore access to the restricted models as soon as possible.
Key Figure Claims Refusal to Cooperate Led to Export Control
David Sacks, co-chair of the President’s Council of Advisers on Science and Technology, posted on X on Saturday that a “highly credible, trusted partner” of both Anthropic and the U.S. government, while testing Fable, discovered a jailbreak of its safeguards. According to Sacks, the administration requested that Anthropic CEO Dario Amodei fix the vulnerability or deactivate the model, but Amodei allegedly refused.
Sacks further claimed that Anthropic’s response contradicted its public statements on AI safety and regulation, a stance recently reiterated by Amodei in a blog post. “Anthropic prioritized the continued offering of the consumer model over safety,” Sacks wrote. He added that the administration issued the export control reluctantly, expressing surprise at Anthropic’s lack of cooperation with a “reasonable safety request.” Sacks characterized Anthropic’s reaction as being “very much at odds with their branding and ethos as a safe AI research community.”
The administration, Sacks noted, hopes Anthropic will resolve the issue, allowing Fable 5 to be re-released. He stated that the administration values Anthropic’s technical capabilities and believes the issue is resolvable, concluding, “The ball is in Anthropic’s court.”
This development follows earlier friction between Anthropic and the U.S. government this year. The AI firm reportedly refused to sign an expanded agreement that would have permitted mass domestic surveillance and the use of fully autonomous lethal weapons systems. Following the breakdown of those talks, President Donald Trump criticized the company, and the Department of Defense designated Anthropic as a “supply chain risk,” a label the firm has challenged legally. More recently, reports suggested a cooling of tensions as the government sought to utilize Claude Mythos and other models.
Leave a Reply