Anthropic has halted access to its Fable 5 and Mythos 5 models, which were launched on June 9. This decision follows a directive from the U.S. government regarding export controls.
The US government, citing national security concerns, 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 employees of Anthropic.
The net effect of…— Anthropic (@AnthropicAI) June 13, 2026
The company stated that the directive formally prohibits access to any foreign nationals, including Anthropic's foreign employees, leading to a complete shutdown of the service.
Anthropic's statement was released on June 12, noting that the directive was received the same day at 5:21 PM Eastern Time. The letter did not specify any particular national security risks.
Anthropic linked the government’s decision to a technique for bypassing Fable 5’s protections. The company claims it reviewed a demonstration and found only a few previously known minor vulnerabilities. According to them, these vulnerabilities are relatively simple, and other publicly available models can also identify them without bypassing protections.
Before the launch of Fable 5, protective mechanisms were tested by the U.S. government, the UK’s AISI, several third-party organizations, and internal teams, totaling thousands of hours of testing. Anthropic stated that a universal jailbreak for the model has not yet been discovered.
The company reminded that Fable 5 has a 30-day data retention policy for clients to investigate and address any bypasses. However, it noted that it had only received verbal information from the government regarding a potential narrow, non-universal jailbreak. The alleged technique, as described by Anthropic, involved asking the model to read a specific codebase and fix software errors.
Anthropic stated that it reviewed the report believed to have informed the directive and concluded that the demonstrated capabilities are also available in other models, including OpenAI's GPT-5.5. The company emphasized that it will comply with the legal requirement but disagrees with the withdrawal of a commercial product due to a narrow potential bypass. In its view, such a standard for the entire industry would effectively halt the launch of new models.
Access to Anthropic's other products remains unchanged. The company apologized to its clients and stated that it is working to resolve what it termed a "misunderstanding."
It is worth noting that in June, Anthropic CEO Dario Amodei urged the U.S. to tighten its approach to regulating artificial intelligence, advocating for mandatory safety checks for the most powerful AI models instead of mere disclosure requirements.