White House Prepares to Give Agencies Access to Powerful Anthropic AI

Concerns raised over cybersecurity risks of Mythos model's capabilities

Apr. 17, 2026 at 2:06am

Got story updates? Submit your updates here. ›

A complex, glowing 3D illustration of futuristic cybersecurity hardware and infrastructure, with neon cyan and magenta lights illuminating the intricate components, conveying the powerful and potentially dangerous capabilities of the Mythos AI model.The government's plan to provide agencies access to Anthropic's advanced Mythos AI raises concerns about the model's potential to expose critical cybersecurity vulnerabilities.Washington Today

The U.S. government is working to make Anthropic's advanced artificial intelligence model, Mythos, available to major federal agencies amid worries that the tool could significantly increase cybersecurity risks. The White House's Office of Management and Budget is setting up protections to allow agencies like the Department of Defense and Department of Homeland Security to start using Mythos, though a timeline is unclear.

Why it matters

Mythos has raised alarms across the government due to its ability to uncover critical software vulnerabilities that would normally require the skills of elite hackers. Officials are grappling with how to evaluate and manage the cybersecurity risks posed by equipping individual actors with such a powerful AI tool.

The details

In an email, the White House's federal chief information officer Gregory Barbaccia told agency technology and cybersecurity leaders to expect more information "in the coming weeks" about accessing Mythos. The model has so far only been provided to a limited group of companies, with Anthropic urging them to use it to assess their own security weaknesses. Anthropic has briefed officials across the government on Mythos' capabilities, including its potential offensive and defensive cyber applications.

  • On April 17, 2026, Barbaccia sent the email to agency officials.
  • In recent months, the U.S. Treasury Department has been seeking access to Mythos to uncover software flaws.

The players

Gregory Barbaccia

Federal chief information officer of the White House Office of Management and Budget.

Anthropic

An artificial intelligence company that developed the powerful Mythos AI model, which has raised cybersecurity concerns across the U.S. government.

U.S. Treasury Department

A government agency that has been seeking access to Mythos to uncover software vulnerabilities.

Got photos? Submit your photos here. ›

What they’re saying

“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.”

— Gregory Barbaccia, Federal chief information officer, White House Office of Management and Budget

What’s next

The White House plans to provide more information to federal agencies in the coming weeks about accessing the Mythos AI model.

The takeaway

The government's move to make Mythos available to agencies highlights the profound cybersecurity risks and uncertainties posed by advanced AI tools that can uncover critical software vulnerabilities. Policymakers must carefully weigh the potential benefits and dangers as they work to manage the national security implications.