Anthropic Suspends Mythos and Fable After Federal Alert

Anthropic recently made the decision to suspend access to its artificial intelligence models, Mythos 5 and Fable 5, following restrictions imposed by the U.S. government. This weekend, the company withdrew these models after federal authorities reported a possible bypass of the security measures put in place by Anthropic. The government informed the company on Friday afternoon that a way to "jailbreak" Fable 5 had been discovered, which would allow users to circumvent the limits designed to reduce the risks of misuse.

At the initial launch of Mythos, Anthropic chose to limit access to a select group of government agencies and technology professionals. This decision was motivated by the software's capabilities to detect cybersecurity vulnerabilities. However, the government imposed export controls on these products, forcing Anthropic to suspend access to both models for any foreign nationals, whether inside or outside the United States. The only solution to comply with these restrictions was to completely shut down the models, according to the company.

Anthropic expressed its disagreement with the government's assessment regarding the jailbreak, stating that it did not warrant such a drastic action. The company warned against the widespread application of this standard, which could "essentially halt" the development of new cutting-edge AI models. Anthropic publicly stated that the government should have the ability to block dangerous deployments, but this should occur within a transparent, fair, clear, and fact-based statutory process. According to Anthropic, the current action does not adhere to these principles.

Representatives from Anthropic and the U.S. Department of Commerce did not immediately respond to requests for comment. However, it has been reported that Anthropic sent staff members to Washington to negotiate with the Trump administration to resolve this issue. The Wall Street Journal indicated that Anthropic's leaders spent several hours on calls Saturday with Commerce Secretary Howard Lutnick and National Cyber Director Sean Cairncross.

This is not the first time Anthropic has found itself in conflict with Washington regarding its AI models. Earlier this year, the Department of Defense labeled Anthropic a "supply chain risk" after the company insisted on restrictions regarding the use of its Claude AI models for fully autonomous weapons or mass domestic surveillance. The Pentagon wanted to be able to use the models for "all legal purposes."