White House Plans To Give Federal Agencies Access To Claude Mythos, The A.I. Model Making Everyone Nervous - HuffPost

The U.S. government is planning to provide federal agencies with access to a version of Anthropic's AI model Mythos, according to a Bloomberg News report. The report states the tool could increase cybersecurity risks. This follows internal warnings from Anthropic experts that Mythos could hack foundational computing systems.

The plan involves making the model available to major federal agencies, though specific agencies were not named in the initial report. The model has been described as making 'everyone nervous' due to its advanced capabilities and potential security implications. Anthropic had previously determined Mythos posed significant risks and restricted its general release.

This development occurs alongside other limited access grants to advanced AI models. BNY Mellon recently obtained early access to both Anthropic's Mythos and OpenAI's GPT-5.4-Cyber for cybersecurity purposes. The federal access plan represents a significant deployment of frontier AI technology within government operations.