Anthropic's Mythos AI Faces US Restrictions Over Security

The U.S. government has ordered Anthropic to suspend access to its Fable 5 and Mythos 5 AI models. This directive comes amid national security concerns. The order requires Anthropic to halt access for foreign nationals, both inside and outside the United States, including foreign national employees at the company. Anthropic has disabled access for all customers to ensure compliance, though other AI systems remain available. This action marks one of the most aggressive government interventions concerning advanced AI models. Anthropic received the directive on June 12 and states the government did not provide specific details on its concerns. However, verbally, officials believe they identified a method to bypass safeguards on Fable 5. Anthropic reviewed the reported technique, finding only minor, known vulnerabilities. They also noted other publicly available models can identify the same issues without bypass methods. The company emphasizes its extensive pre-launch testing for Fable 5, which showed stronger safeguards than previous models. The only evidence received from the government involves a narrow "jailbreak" related to the model reviewing code for software flaws. This capability, Anthropic states, is common in the industry and used by cybersecurity professionals. This situation highlights the ongoing tension between AI development and national security oversight.