OMB’s examination of Mythos is ‘not giving access to anything to agencies,’ official says
Celal Gunes/Anadolu via Getty Images
Following the debut of Anthropic’s ultra-powerful AI, Mythos, the federal CIO told stakeholders the administration is meeting with industry and intelligence partners to discuss guardrails and safeguards associated with “potentially releasing a modified version of the model.”
The Office of Management and Budget isn’t currently planning any policy changes related to granting federal agencies access to Anthropic’s powerful Mythos artificial intelligence model, despite an email from the federal chief information officer, Gregory Barbaccia, indicating the administration is looking into the technology.
Bloomberg reported Thursday that the White House is working to give agencies access to Mythos — a new high-power AI model that has drawn widespread attention for identifying a tidal wave of digital vulnerabilities — though an OMB spokesperson clarified to Nextgov/FCW that, as of now, there “are no policy changes and there is no OMB policy process happening on this issue.”
“OMB is not giving access to anything to agencies,” the spokesperson told Nextgov/FCW on Friday.
Mythos was made available to a select set of industry partners under Anthropic’s Project Glasswing, but news of its capabilities has sparked a flurry of meetings between government, AI companies and industry stakeholders. The Wall Street Journal reported last week that National Cyber Director Sean Cairncross is convening meetings to prepare for the cyber repercussions of Mythos and anticipated powerful AI systems from other companies.
Earlier this week, Jack Clark, Anthropic’s co-founder and head of Public Benefit, said there will be other systems like Mythos in the near future and that the world needs to “get ready” for more powerful AI.
Barbaccia’s email told stakeholders the administration is doing just that.
“Thank you to everyone who has reached out to proactively work with us on this important initiative,” the email, first obtained by Bloomberg and later seen by Nextgov/FCW, said. “We're working closely with model providers, other industry partners, and the intelligence community to ensure the appropriate guardrails and safeguards are in place before potentially releasing a modified version of the model to agencies.”
Barbaccia went on to say that OMB and the Office of the National Cyber Director will continue updating relevant parties and are expected to have more information on Mythos and its potential role in government “in the coming weeks,” but did not promise that agencies will get access to the model.
Mythos’ debut also comes as Anthropic and the Trump administration continue litigation over the Pentagon labeling the company a supply chain risk after it refused to allow its systems to be used in autonomous lethal weapons and in the surveillance of Americans.
President Donald Trump subsequently ordered federal agencies to offload Anthropic’s tools from their networks within a six-month timeframe. The designation of the U.S. company as a supply chain risk and the order to terminate government contracts has spurred anxiety among the vendor community over how future business with the government will be conducted.
On Friday, Politico reported plans for Anthropic CEO Dario Amodei to meet with White House Chief of Staff Susie Wiles, along with other Trump administration officials.