Together pentagon And treasury departmentAt least three other agencies have reportedly stopped using Anthropic products, including state DepartmentThe Department of Health and Human Services and this federal housing finance agency.
Following divided judicial decisions in lawsuits brought by Anthropic in California and Washington DC, some government offices are already avoiding the government supply chain risk designation and evaluating Mythos’ impact on cybersecurity. And agencies like the Treasury Department – which recently sought access Mythos – one of many that are now seeking guidance on how to proceed – has been asked to determine its implications for the cyber security of banking firms.
The Office of Management and Budget is now considering whether agencies will be allowed to use a “revised” version of Mythos, according to an email sent to federal agencies and obtained by POLITICO.
“We are working closely with model providers, other industry partners, and the intelligence community to ensure appropriate guardrails and safeguards are in place before agencies potentially release a revised version of the model,” OMB Chief Information Officer Gregory Barbaccia wrote to several federal agencies earlier this week. Barbaccia said OMB and the Office of the National Cyber Director “will continue to keep everyone updated and expect more information in the coming weeks.”
National cyber director Sean Cairncross convened a call this week with representatives from private sector companies to discuss Mythos, according to a participant on the call, who asked to remain anonymous because they were not authorized to speak publicly about it.
Cairncross said ONCD has the support of Wills and Vice President J.D. Vance to lead the Trump administration’s response to the hacking capabilities of Anthropic’s powerful new model, the person said. During the roughly ten-minute call, Cairncross both praised Mythos as a sign of US AI innovation and warned companies about the risks it presents to their networks.
A White House official, who declined to be named because they were not authorized to speak on the record, told POLITICO that it is actively engaging with government and industry, including Frontier AI labs whose models require technical assessment and can help secure critical U.S. software vulnerabilities. “The collective effort of everyone involved will ultimately benefit the industry and our country,” the person said.
OMB spokesperson, ONCD, Treasury and the Office of the Vice President did not immediately respond to POLITICO’s requests for comment. A spokesperson for Anthropic declined to comment ahead of today’s meeting.
John Hewitt Jones contributed to this report.
