OpenAI and Google Employees Filed a Brief Defending Anthropic Against the Pentagon
March 29, 2026 · 6 min read
TL;DR
The Pentagon blacklisted Anthropic as a “supply-chain risk” after Claude refused contract language authorizing domestic mass surveillance and autonomous weapons use. Anthropic filed suit on March 9, 2026. The next day, 30+ OpenAI and Google DeepMind employees — including Google chief scientist Jeff Dean— filed an amicus brief in Anthropic's defense. Microsoft filed its own separate brief. Multiple OpenAI employees resigned over OpenAI's own Department of War contract. The case is ongoing. No ruling as of March 29, 2026.
What happened: the dispute in plain English
The Pentagon wanted a contract with Anthropic that included a clause authorizing “all lawful use” of Claude AI systems. Anthropic interpreted this as covering domestic mass surveillance operations and autonomous weapons systems — uses that Anthropic's AI safety and ethics policies explicitly prohibit.
Anthropic refused to agree to the clause. The Department of Defense responded by designating Anthropic a “supply-chain risk” — a national security designation that blocks all federal agencies from purchasing or using Anthropic products. Anthropic filed a lawsuit on March 9, 2026 challenging the designation as unlawful government retaliation.
The scale of industry solidarity that followed was unusual. AI companies are fierce competitors. Jeff Dean — Google DeepMind's chief scientist, a person whose work directly competes with Anthropic's — signed a legal brief supporting Anthropic hours after the lawsuit was filed. The employees' argument: the precedent set by blacklisting Anthropic threatens every company in the industry.
Case timeline
Late 2025 – Early 2026
—Pentagon negotiations break down
The DOD requests a contract clause authorizing 'all lawful use' of Claude, including domestic surveillance and autonomous weapons. Anthropic refuses on AI safety and ethics grounds.
March 9, 2026
—Anthropic files lawsuit
Anthropic files suit in federal court challenging the Pentagon's 'supply-chain risk' designation as unlawful, arguing the designation bypasses standard procurement processes and constitutes government retaliation.
March 10, 2026
—Industry amicus briefs filed
30+ OpenAI and Google DeepMind employees file an amicus brief including Google chief scientist Jeff Dean. Microsoft files its own separate brief. Both argue the blacklist threatens U.S. AI competitiveness.
March 10–15, 2026
—OpenAI employee resignations
Multiple OpenAI employees resign over OpenAI's separate contract with the Department of War, citing concerns about military AI applications. OpenAI's relationship with DOD becomes a source of internal tension.
March 15, 2026
—Anthropic Institute announced
Anthropic announces the Anthropic Institute, a new research entity focused on the economic, societal, and security impacts of advanced AI. Widely seen as a move to establish an independent research voice separate from government pressure.
Ongoing
—Case proceeds in federal court
The lawsuit is proceeding. No ruling as of March 29, 2026. An earlier court ruling in Anthropic's favor on procedural grounds (covered separately) established initial favorable precedent.
The parties and their positions
| Party | Role | Position |
|---|---|---|
| Anthropic | Plaintiff | Refused DOD 'all lawful use' clause. Filed lawsuit March 9 challenging blacklist as unlawful government retaliation. |
| U.S. Department of Defense | Defendant | Blacklisted Anthropic as supply-chain risk after failed contract negotiations. Maintains designation is within authority. |
| 30+ OpenAI / Google DeepMind employees | Amicus (supporting Anthropic) | Filed brief arguing blacklist threatens U.S. AI industry competitiveness. Includes Google chief scientist Jeff Dean. |
| Microsoft | Amicus (supporting Anthropic) | Filed separate brief. Business interest in protecting AI vendors from retaliatory procurement actions. |
| OpenAI (company) | Neutral / conflicted | Company itself did not file — has separate DOD contract. Multiple employees resigned over military AI contracts. |
Why this matters beyond the lawsuit
The Anthropic Pentagon dispute is the clearest test to date of whether AI companies can enforce ethical use policies against government clients. The outcome will define the leverage that any government has over AI vendors who refuse to enable specific use cases.
If Anthropic wins: it establishes that retaliatory blacklisting of an AI company for ethical refusals is unlawful. Other AI companies gain legal precedent protecting their right to decline government contract clauses that violate their use policies.
If the Pentagon wins: every AI company operating in the U.S. faces a credible threat of supply-chain blacklisting if it refuses government use cases — regardless of those companies' safety or ethical commitments. The AI industry loses the ability to enforce its own policies against its most powerful customers.
Frequently asked questions
Why did the Pentagon blacklist Anthropic?
The Pentagon blacklisted Anthropic in early March 2026 as a 'supply-chain risk' after Anthropic refused to agree to Pentagon demands for 'all lawful use' of Claude AI systems. The disputed clause would have authorized the Department of Defense to use Claude for domestic mass surveillance and autonomous weapons systems. Anthropic's position is that these uses violate its AI safety and ethics policies. The Defense Department's response was to designate Anthropic a national security supply-chain risk — a designation that blocks federal agencies from using Anthropic's products. Anthropic filed a lawsuit on March 9, 2026 challenging the designation as unlawful.
What did the OpenAI and Google employees actually sign?
More than 30 employees from OpenAI and Google DeepMind signed an amicus brief (a formal legal filing by parties not directly involved in the lawsuit who have a relevant interest in the outcome) in Anthropic's lawsuit against the Pentagon, filed on March 10, 2026. The signatories included Google DeepMind chief scientist Jeff Dean — the highest-profile cross-company signatory. The brief argued that the Pentagon's blacklisting of Anthropic threatens to damage the entire American AI industry by establishing a precedent that AI companies can be punished for refusing to enable specific military use cases. Microsoft filed a separate amicus brief in support of Anthropic the same day. Multiple OpenAI employees also reportedly resigned over OpenAI's separate contract with the Department of War.
What does this mean for users of Claude AI?
For current Claude users, the Pentagon dispute has no immediate impact on Claude's availability, pricing, or capabilities. Anthropic's commercial operations continue normally. The lawsuit concerns U.S. federal government procurement, not consumer or business product access. The longer-term implications: if Anthropic prevails, it establishes legal precedent that AI companies cannot be compelled to enable specific military use cases through supply-chain risk designations. If the Pentagon prevails, AI companies face a credible threat of blacklisting if they refuse government-mandated use cases — which could create pressure on all AI labs to accept broader government use agreements. For Anthropic specifically, the ruling will affect its ability to win future federal government contracts.
Why did OpenAI and Google employees support a competitor?
The OpenAI and Google employees who signed the amicus brief were acting on shared industry interest rather than supporting a competitor's business. Their stated reasoning: the precedent set by blacklisting Anthropic threatens every AI company that refuses government demands. If federal agencies can blacklist AI labs for declining specific use cases, every AI company faces the same risk — including OpenAI and Google. Jeff Dean and the other signatories argued that undermining a leading U.S. AI company on safety grounds damages American AI competitiveness globally. The move is unusual but not unprecedented in the tech industry, where employees from competing companies have occasionally united on legal or regulatory issues that threaten the sector as a whole.
Stay current on AI policy and industry developments
Happycapy + Capymail delivers weekly AI industry intelligence to your inbox — policy updates, company moves, model releases — without requiring you to follow dozens of news sources. $17/month. Free tier available.
Try Happycapy Free →