The US government is preparing to make a version of Anthropic PBC's powerful new artificial intelligence model available to major federal agencies amid concerns that the tool could sharply increase cybersecurity risk, according to a memo reviewed by Bloomberg News.

Gregory Barbaccia, federal chief information officer of the White House Office of Management and Budget, told officials at Cabinet departments in an email Tuesday that OMB is setting up protections that would allow their agencies to begin using the closely guarded AI tool, Mythos.

The email doesn't say definitively that the various agencies will get access to Mythos, nor does it provide a timeline for when it might come or how they might use it. It tells top technology and cybersecurity chiefs to expect more information "in the coming weeks."

Anthropic has only provided Mythos to a limited group of technology companies, financial firms and others, urging them to use it to assess their cybersecurity risk. The firm limited the release of Mythos amid concerns that hackers could weaponize its capabilities to steal data or sabotage victim networks.

Get the Tech Newsletter bundle.

Bloomberg's subscriber-only tech newsletters, and full access to all the articles they feature.

Plus Signed UpPlus Sign UpPlus Sign Up

By continuing, I agree to the Privacy Policy and Terms of Service.

Before its limited release of Mythos, Anthropic briefed senior officials across the US government on the model's full capabilities, including both its offensive and defensive cyber applications, according to a company official who spoke on condition that they not be identified discussing the talks with government. The talks included staff at the Cybersecurity and Infrastructure Security Agency and the Center for AI Standards and Innovation, among others, the company official said, and Anthropic has continued to work with government on security issues arising from the model.

Barbaccia's message was sent as leaders from Washington to Wall Street are grappling with the possibility that the model could make it dramatically easier for hackers to find ways to break into sensitive computer systems in industry and government.

"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 wrote in the email, which had the subject, "Mythos Model Access."

A White House official said in an email that the government continues to work and engage with AI companies to ensure their models help secure critical software vulnerabilities. They didn't answer specific questions on the matter.

Anthropic declined to comment.

Neither Anthropic nor the government said what, if any, federal agencies have gotten early access to Mythos.

Barbaccia's email went to officials with the Department of Defense, Department of Treasury, Department of Commerce, Department of Homeland Security, Department of Justice and Department of State, among several other agencies.

The Treasury Department has been seeking access to Mythos in order to uncover its own software flaws, Bloomberg has reported.

Within Anthropic, company leaders became worried the model could be a national security risk after testers were able to use Mythos to turn up the types of critical bugs that it would normally take the world's best hackers to uncover. These concerns prompted the company's limited release of the model.

It's similarly set off alarms in various parts of the US government.

Among officials focused on national defense, the introduction of Mythos has created profound uncertainty about how to evaluate cybersecurity risk, a person familiar with the matter previously told Bloomberg. Equipping an individual hacker with the model, or similar AI tools, would likely be a transformation equivalent to turning a conventional soldier into a special forces operator, the person said.

On the day, Anthropic publicly disclosed Mythos' existence, US Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell convened Wall Street leaders for a meeting in Washington to urge them to use the model to find weaknesses in their own systems.

White House Moves to Give US Agencies Anthropic Mythos Access