Editorial illustration for Anthropic may keep supplying Claude to NSA despite Pentagon risk flag
Anthropic may keep supplying Claude to NSA despite...
The Pentagon flagged Anthropic as a supply chain threat. The NSA needs chips it doesn’t have. And yet, a deal is nearly done.
Anthropic’s new “Mythos” model runs on older hardware, a rare commodity in an intelligence world starved of Nvidia’s latest Grace Blackwell chips. That technical quirk makes it the only viable short-term fix for the NSA’s classified networks. The contract includes a clause barring the model from processing Americans’ data.
The earlier, deal-breaking demand for “any lawful use” is gone. The White House wants this agreement to become a template. It has also approved $9 billion for new AI chips to solve the shortage long-term.
For now, the NSA gets its AI fix from a company the Pentagon itself flagged as a supply chain risk.
Anthropic, which the US government has labeled a "supply chain risk," will likely keep delivering AI models to the NSA anyway. White House Chief of Staff Susie Wiles personally approved the arrangement, the New York Times reports. Intelligence agencies don't have enough of Nvidia's latest Grace Blackwell chips to run current models from OpenAI and others.
The White House has found its template. And in that template, one contradiction sits at the center: the Pentagon flagged Anthropic as a risk, yet the NSA needs its model to fill a chip-shaped hole. That hole is temporary , $9 billion will eventually close it , but for now, the logic is brutal.
Better a flagged contractor than an intelligence gap. The clause barring the model from processing Americans’ data offers a narrow guardrail, not a moral resolution. It signals a new bargain: national security buys what it can, when it can, and worries about the supply chain later.
What remains unsettled is the precedent. Once you enshrine expedience over principle in a government-wide template, the exception becomes the rule. Anthropic supplies the NSA.
The Pentagon objects. The White House funds the fix. And the next company in line learns that the fastest way past a red flag is to be the only answer in the room.
Related Reading
Browse all Anthropic news →
LLMs & Generative AI ChatGPT's 'Nerdy' tweak rewards goblin metaphors in answers, study finds
LLMs & Generative AI Google tests visual 'magazine-style' UI for Gemini 3 Pro users
LLMs & Generative AI AI Engineers Face Rising Costs, Need New Strategies for Efficiency
Related Topic Trump cracks down on Anthropic after Amazon tip; staff largely foreign
Open Source
Related Topic Claude Mythos highlights EU AI safety gaps, says researcher Caroli
Policy & Regulation
From Archives Claude Code auto‑creates AI scaling algorithms; new control allocates compute
From Archives SuperClaude workflow ranks security issues, details attack vectors, gives fixes
From Archives Anthropic: Claude Mythos Preview finds ~3,900 high‑severity open‑source bugs
From Archives Alibaba's Qwen3.7-Max runs 35 hrs, self‑monitors reward‑hacking, supports Claude Code
Common Questions Answered
Why is Anthropic's Mythos model considered viable for the NSA despite Pentagon concerns?
Anthropic's Mythos model runs on older hardware, making it compatible with systems that don't require Nvidia's latest Grace Blackwell chips, which are in severe shortage across the intelligence community. This technical capability makes it the only practical short-term solution for the NSA's classified networks, outweighing the Pentagon's earlier supply chain risk flag.
What safeguard does the Anthropic-NSA contract include to protect Americans' privacy?
The contract includes a specific clause that bars Anthropic's Mythos model from processing Americans' data on classified networks. This restriction serves as a guardrail to limit potential privacy risks, though experts note it represents a narrow technical constraint rather than a comprehensive policy solution.
How does the Pentagon's supply chain threat flag conflict with the NSA's need for Anthropic's technology?
The Pentagon flagged Anthropic as a supply chain risk, yet the NSA still needs its Mythos model to address a critical chip shortage on its classified networks. This contradiction reflects a pragmatic decision that prioritizes filling an immediate intelligence capability gap over supply chain concerns, with the logic being that avoiding an intelligence gap outweighs the flagged contractor risk.
When will the NSA's chip shortage be resolved according to the article?
The article indicates that the chip shortage is temporary and will eventually be closed by a $9 billion investment. However, this long-term solution is not yet available, making Anthropic's Mythos model necessary as a short-term fix for the NSA's current classified network needs.
Further Reading
- Anthropic says Pentagon risk flag won’t stop Claude from being used by federal customers outside DoD work — Anthropic
- Pentagon Designates Anthropic a Supply Chain Risk - What Government Contractors Need to Know — Mayer Brown
- Pentagon designates Anthropic a supply chain risk — Reuters via YouTube
- What IT Leaders Should Know About AI Supply Chain Risk Ruling — Janus Associates
- Pentagon Officially Flags Anthropic as Supply Chain Risk — Deeper Insights
