Anthropic Fable 5 suspended alongside Mythos 5 after Washington issues a surprise export control directive over national security concerns
Anthropic has disabled access to two of its most advanced AI models following a directive from the US government. Fable 5 and Mythos 5 are now unavailable to all customers after Washington issued an export control order citing national security authorities.
The order requires Anthropic to suspend access to both models for all foreign nationals. This applies whether those individuals are inside or outside the United States. Moreover, the restriction extends to foreign national employees working at Anthropic itself.
To ensure compliance, Anthropic abruptly disabled Fable 5 and Mythos 5 for its entire customer base. Access to all other Claude models remains unaffected. The company apologized to customers for the disruption and said it believes the situation stems from a misunderstanding. Furthermore, Anthropic confirmed it is working to restore access as quickly as possible.
The government contacted Anthropic on June 12 at 5:21 pm ET. However, the directive provided no specific details about the national security concern it cited. Anthropic said its understanding is that the government believes someone discovered a method to jailbreak Fable 5. The company reviewed a demonstration of the technique and found it identified a small number of previously known and minor software vulnerabilities. Additionally, Anthropic noted that the same vulnerabilities were discoverable using other publicly available AI models without the bypass.
The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.
— Anthropic (@AnthropicAI) June 13, 2026
The net effect of…
Anthropic launched Claude Fable 5 on June 9 as a Mythos-class model made safe for general use. The company described it as its most capable publicly available model, with strong performance across software engineering, knowledge work, vision, scientific research, and long-running tasks. Meanwhile, Mythos 5 launched for a smaller group of cyber defenders and infrastructure providers through Project Glasswing. It shares the same underlying model as Fable 5 but operates with certain safeguards lifted for trusted users.
Before launching Fable 5, Anthropic worked with the US government, the UK AI Security Institute, third-party organizations, and internal teams to test the model’s safeguards for thousands of hours. No tester found a universal jailbreak capable of broadly bypassing those safeguards. Furthermore, Anthropic uses a defense-in-depth approach, aiming to make any jailbreak narrow and costly to execute while monitoring for successful attacks.
Senior Anthropic technical staff are scheduled to meet US Department of Commerce officials in Washington to discuss the dispute. Reuters reported that Anthropic staff had already held virtual meetings with government officials every day since receiving the directive.
Cybersecurity experts have also pushed back against the restrictions. A group of security leaders argued that blocking access to Anthropic’s most advanced models could hurt defenders more than attackers. They noted that similar capabilities exist across other leading AI models. Therefore, restricting Fable 5 specifically could create market uncertainty without meaningfully reducing broader security risks.
Anthropic said it complies with the legal directive but strongly disagrees with the reasoning behind it. The company argued that a narrow potential jailbreak should not trigger the recall of a commercial model serving a large number of users. Moreover, applying the same standard across the AI industry could effectively halt new model deployments from frontier providers entirely.
Anthropic concluded that the government should retain authority to block unsafe deployments. However, it insisted that any such action must follow a transparent and fair process grounded in technical facts. In its view, this directive does not meet that standard.
