OpenAI, DeepMind staff back Anthropic in Pentagon risk dispute

More than 30 OpenAI and Google DeepMind employees filed an amicus brief supporting Anthropic’s lawsuit after the US Department of Defense labeled Anthropic a supply-chain risk.

Jeff Dean, Google DeepMind chief scientist, signed the brief.

The Pentagon labeled Anthropic a supply-chain risk after Anthropic refused to allow the DOD to use its technology to conduct mass surveillance of Americans or to autonomously fire weapons.

The DOD said it should be able to use AI for any “lawful” purpose and not be constrained by a private contractor.

Anthropic filed two lawsuits, and the amicus brief appeared on the docket a few hours later.

Signers warned the designation could harm US competitiveness in AI and said contractual and technical limits are important safeguards while public law remains unsettled.

The DOD signed a deal with OpenAI within moments of the designation, a move that many OpenAI employees protested.

🔗 Source: TechCrunch

🧠 Food for thought

Implications, context, and why it matters.

The Pentagon used a law meant for foreign adversaries against an American AI company

The “supply chain risk” label likely leans on 10 U.S.C. § 3252, a statute long used to block tech tied to foreign adversaries rather than handle contract fights with U.S. firms 1.
This appears to be the first time the government has used this designation on a U.S. company 2.
The Defense Secretary’s public order appears to stretch the law by trying to stop military contractors from doing any commercial business with Anthropic, not only government work 1.
Public comments that called Anthropic a “radical left, woke company” add weight to Anthropic’s claim that the action was meant as punishment, a core part of its lawsuit alleging a free speech violation 2.

The Pentagon’s move is spreading through Anthropic’s business and a wider AI ethics fight

The designation is already hitting private deals, with a financial services client pausing a $15 million contract and a drugmaker seeking a shorter term 3.
Anthropic has spent over $10 billion training its models and said the uncertainty threatens its ability to raise capital 3.
The administration is drafting procurement rules for civilian AI contracts that would require vendors to give the government an irrevocable license to use their models for any legal purpose 4.
The clash pushes private tech companies toward setting policy for AI-powered surveillance and autonomous weapons, after Congress left a gap 5.