CISA Locked Out of Anthropic's Mythos Preview

Does CISA have access to Anthropic's Mythos Preview?

The Cybersecurity and Infrastructure Security Agency (CISA) does not have access to Anthropic's Mythos Preview model. Two sources told Axios this directly. More than 40 other organizations already have access to the model.

This matters because CISA is the country's top cyber defense agency. It helps secure banks, power plants, and other critical infrastructure. Those industries are increasingly worried about AI-powered cyberattacks overwhelming their defenses.

Why doesn't CISA have access?

The sources don't give a single explicit reason. What is clear: CISA is not on Anthropic's current access list, even though some other government agencies are using Mythos Preview.

Earlier this month, an Anthropic official told Axios the company had briefed CISA and the Commerce Department on Mythos' capabilities. A briefing is not the same as access.

The Trump administration has also spent the past year reducing capacity at CISA. It has shifted more policy influence to the White House's national cyber director and pushed some programs to the state and local level.

You might also like

SpaceX Joins Nasdaq 100 July 7 With $27B Buy

De Beers Cuts Diamond Prices to Near Market Rate

China Confirms US Farm Tariff Relief Coming

Top US Universities Slash PhD Admissions in 2026

Who is negotiating to change this?

National cyber director Sean Cairncross is among the Trump officials currently negotiating broader civilian agency access to Mythos, according to Axios.

Here's what we know so far: those negotiations are ongoing, and no deal has been reported as finalized.

What is Mythos Preview, and who has it?

Mythos Preview is Anthropic's powerful new AI model currently in limited release. More than 40 organizations have been granted access. Some of those are other government agencies — but CISA is not among them.

Separately, Nextgov/FCW has reported that parts of the NSA lost access to Mythos 5 amid an Anthropic supply chain dispute — a sign that government access to Anthropic models is uneven and contested across multiple agencies.

What does this mean for critical infrastructure security?

Security teams at critical infrastructure organizations have historically relied on CISA to share threat intelligence across sectors. They also look to CISA to help prioritize security strategies.

Without access to Mythos Preview, CISA cannot directly evaluate the model's capabilities or use it for that threat-intelligence work. The industries it serves are already concerned about AI-powered attacks.

This situation connects to broader questions about how AI tools reach US government defenders — a topic that has also surfaced in Anthropic's Pentagon disputes and in efforts like the Palantir-NVIDIA government AI initiative.

Key facts at a glance

What happens next?

Sean Cairncross is actively negotiating broader civilian agency access to Mythos. That process is ongoing. No timeline or outcome has been confirmed in the sources.

CISA's acting director, Nick Andersen, appeared at a congressional hearing recently, though the sources do not detail what was said there about Mythos access specifically.

For builders and developers watching how AI tools move through government channels, this situation sits alongside other access and procurement friction — similar to the dynamics covered in our reporting on Claude's government use and sovereign government AI deployments.