Anthropic blocked public access to its Fable 5 and Mythos 5 AI models on Friday, hours after the Trump administration imposed new export restrictions on the systems over national security concerns. The White House acted "partly over suspicions that a China-linked group had accessed" the Mythos 5 model, a person familiar with the matter told the Washington Examiner.
The export controls effectively bar foreign entities from obtaining the models without US government approval, escalating an ongoing policy push to limit the spread of advanced AI technology. The move targets two of Anthropic's most capable releases, suggesting the administration views them as potential dual-use tools that could aid adversaries.
The restrictions underscore deepening bipartisan consensus in Washington that frontier AI models pose a strategic risk. While the White House has not formally identified the suspected group, the allegation of China-linked access is likely to intensify calls from Republican lawmakers for tighter export licensing and mandatory security audits of AI companies.
Public reaction has been muted so far, but industry observers note the action could set a precedent for how the US regulates open-weight models. Some civil liberties groups have warned that broad restrictions may stifle academic research and global collaboration without proving effective at stopping determined state actors.
Counter argument: Critics argue that the evidence of Chinese access remains unverified and that restricting public access to AI models disproportionately harms legitimate researchers and startups. They contend the administration has not demonstrated that export limits are more effective than existing cybersecurity measures at preventing unauthorized use.