NSA to Use Anthropic's Claude Despite Pentagon Supply Chain Ban

Key Takeaways

• White House Chief of Staff Susie Wiles personally approved Anthropic's NSA arrangement despite Pentagon blacklist
• Anthropic's Mythos model runs on older chips, solving the NSA's hardware shortage problem
• The contract bars processing Americans' data and drops controversial 'any lawful use' language

Anthropic will likely continue supplying its Claude AI models to the National Security Agency. This happens despite the Pentagon officially labeling the company a national security supply chain risk in February 2026. White House Chief of Staff Susie Wiles personally signed off on the arrangement, according to the New York Times.

The contradiction is stark. One arm of the federal government has blacklisted Anthropic from procurement. Another arm has quietly approved a contract to use Anthropic's technology in classified intelligence networks.

Why the NSA Needs Anthropic Right Now

The answer comes down to hardware. Intelligence agencies lack enough of Nvidia's latest Grace Blackwell chips to run current AI models from OpenAI and other providers. Anthropic's new 'Mythos' model solves this problem. It runs on older chips that the NSA actually has deployed across its classified networks.

According to the Times report, Mythos is currently the only short-term fix for the NSA's AI capabilities gap. The agencies need functional AI now. They cannot wait for new chips to arrive and be installed across secure facilities.

The Supply Chain Risk Designation

The Pentagon classified Anthropic as a supply chain threat for a specific reason. The company refused to release its technology for 'any lawful use.' Anthropic has maintained safety guardrails that prohibit autonomous weapons and certain surveillance applications. The Defense Department interpreted this as an unacceptable restriction on government flexibility.

“The designation of Anthropic as a supply chain risk is unprecedented, unlawful, and a clear attempt at retaliation for our refusal to abandon our safety principles.”

— Anthropic Spokesperson, official statement regarding March 2026 lawsuit against the DoD

Anthropic filed suit against the Department of Defense in March 2026, challenging the designation as both unprecedented and retaliatory. That litigation remains active. The company is now simultaneously suing the Pentagon while supplying AI to the NSA under White House authorization.

What the Contract Actually Says

The contract being finalized includes specific guardrails. It bars the Mythos model from processing Americans' data. The controversial 'any lawful use' language that derailed earlier negotiations is not part of this deal.

The White House plans to use this contract as a template for future agreements with other AI companies. If successful, it could establish a precedent for how the government procures AI technology from companies that maintain safety restrictions.

The $9 Billion Long-Term Fix

Separately, the White House has approved $9 billion for new AI chips to address the hardware shortage over time. Congress still needs to approve the funding. The money would go toward building specialized data centers to reduce reliance on third-party AI models.

Until those data centers come online, Anthropic's Mythos model remains the practical solution. The NSA cannot pause its AI capabilities while waiting for Nvidia's latest chips to arrive.

The Precedent Problem

This arrangement raises questions about what a supply chain risk designation actually means. If the White House can override Pentagon security designations based on tactical needs, the designation becomes advisory rather than binding.

Discussion across Hacker News and Reddit has been split. Some view this as a necessary pragmatic compromise. Hardware shortages are real. The NSA needs AI capabilities. Anthropic's model works.

Others see a dangerous precedent. The executive branch is now overriding official national security classifications. The primary AI supplier is actively suing the Pentagon. These are not normal procurement conditions.

Current Access Restrictions

The Mythos model preview is currently available to 40 vetted organizations. Anthropic's safety contracts with federal agencies permit zero autonomous weapons system use cases. These restrictions appear to remain in place under the new NSA arrangement.

The arrangement shows how chip shortages have shifted leverage in AI procurement. Anthropic refused to drop its safety restrictions. The government designated it a supply chain risk. But the government still needs Anthropic's technology because it is the only option that works with existing hardware.

Frequently Asked Questions

Why did the Pentagon designate Anthropic a supply chain risk?

Anthropic refused to release its AI technology for 'any lawful use.' The company maintains safety guardrails prohibiting autonomous weapons and certain surveillance applications. The Pentagon viewed these restrictions as unacceptable limitations on government flexibility.

What makes Anthropic's Mythos model different from other AI models?

Mythos runs on older chips that intelligence agencies already have deployed. Other current AI models require Nvidia's latest Grace Blackwell chips, which the NSA lacks in sufficient quantities.

Does the NSA contract allow processing of Americans' data?

No. The contract being finalized explicitly bars the Mythos model from processing Americans' data.

Is Anthropic still suing the Pentagon?

Yes. Anthropic filed suit against the Department of Defense in March 2026, challenging its supply chain risk designation as 'unprecedented, unlawful, and retaliatory.' The litigation remains active.

How will the chip shortage be resolved long-term?

The White House has approved $9 billion for new AI chips and specialized data centers. Congress must still approve the funding.

Source: The Decoder / Matthias Bastian

Preparations for Mythos Public Preview and Security Dashboard Expansion

The new article indicates that Anthropic is preparing for a potential public release of the Mythos model, evidenced by the discovery of 'claude-mythos-1-preview' references in source code and cloud infrastructure. It also reveals that Anthropic is developing a revamped Claude Security dashboard to display vulnerability discovery metrics, expanding on the model's capabilities and its path toward broader access.