The AI company behind Claude is fighting back after being labeled a national security threat for refusing to let its chatbot be used in autonomous weapons and mass surveillance. The stakes couldn’t be higher.
Somewhere between a Silicon Valley boardroom and the Pentagon, a line got drawn in the sand. On one side: Anthropic, the AI safety company behind Claude. On the other: the Trump administration, which labeled that company a national security risk for daring to say its technology should not be used to spy on Americans or autonomously kill people.
On Monday, Anthropic filed back. Two federal lawsuits. Dozens of defendants. And a 48-page argument that the U.S. government just did something it has never done before to an American company.
“The Constitution does not allow the government to wield its enormous power to punish a company for its protected speech.” — Anthropic lawsuit filing, March 9, 2026
Welcome to the most consequential legal battle in the history of artificial intelligence.
How We Got Here: The Backstory
For the past year, Claude has been operating inside some of the most sensitive systems in the U.S. government. Not just bureaucratic email drafts. Actual classified networks. According to reporting from the Wall Street Journal, Claude has reportedly been used in military operations, including intelligence assessments and target identification in ongoing conflicts. And as Mayer Brown’s legal analysis notes, a July 2025 contract made Claude the first frontier AI model approved for use on classified government networks.
Anthropic knew this. They supported it. What they would not support was handing over the keys with zero conditions attached.
The Pentagon wanted a new contract. One that gave the military the right to use Claude for, in their words, “all lawful use.” Anthropic drew two bright lines in response: no mass surveillance of U.S. citizens, and no use in autonomous weapons systems without meaningful human oversight.
From Anthropic’s perspective, this was not a political stance. CEO Dario Amodei publicly explained that Claude simply cannot be reliably and safely deployed for these use cases at the current stage of the technology. That is not ideological posturing. That is an honest technical assessment.
The Pentagon saw it differently.
The Moment Everything Broke Down
Amodei met personally with Defense Secretary Pete Hegseth on February 24th. No agreement was reached. Four days later, on February 27th, President Trump posted on Truth Social:
“WE will decide the fate of our Country — NOT some out-of-control, Radical Left AI company run by people who have no idea what the real World is all about.”
That same day, federal agencies were ordered to immediately stop using Anthropic’s technology.
Then came March 5th. Hegseth formally issued something that has never before been applied to an American company: the “supply chain risk to national security” designation. Under this label, any defense contractor doing business with the Pentagon must certify they are not using Anthropic’s AI.
Lockheed Martin began cutting ties. Other contractors followed. Anthropic’s government-related revenue, and potentially billions in broader business, started evaporating overnight.
Meanwhile, just hours after the Anthropic-Pentagon talks collapsed, OpenAI quietly signed its own deal with the Defense Department. No stated restrictions. A move that Amodei later, in an internal memo reported by The Information, described as “gullible” on the part of OpenAI staff, accusing their leadership of spreading misinformation about the situation. OpenAI later acknowledged the announcement looked “sloppy and opportunistic” and said it was renegotiating some of its terms. Relations between the two AI giants have since deteriorated sharply.
What Anthropic Is Actually Arguing in Court
Anthropic filed two lawsuits simultaneously on Monday. One in the U.S. District Court for the Northern District of California — the detailed, 48-page case. Another in the D.C. Circuit Court of Appeals, targeting a separate statute the government invoked. As Lawfare explains, the five-count complaint follows weeks of tension over Anthropic’s two ethical red lines.
Together, they make three core arguments:
- First Amendment retaliation: The designation punishes Anthropic for expressing a protected viewpoint on AI policy. The government cannot blacklist a company for its speech.
- Statutory overreach: The supply chain risk law (10 U.S.C. 3252) requires the Pentagon to use the least restrictive means available. Anthropic argues a simpler option existed: terminate the contract and hire a different developer. Instead, they chose to make Anthropic radioactive to the entire defense contracting ecosystem.
- No congressional authority: As Just Security’s legal analysis argues, the designation “does not give the Secretary the power to prohibit defense contractors from engaging in any commercial activity” with a designated company. The powers granted by Congress are far narrower than what the Pentagon attempted here.
The lawsuit also includes something notable in its exhibits: screenshots of Trump’s Truth Social posts and Hegseth’s statements on X, submitted as evidence of retaliatory intent. Legal experts writing in Lawfare have already concluded the designation “exceeds what the statute authorizes” and that Hegseth’s own public statements “may have doomed the government’s litigation posture before it even begins.”
The Wildcard: Who’s Fighting With Anthropic
Here is the plot twist that no one saw coming.
The law firm representing Anthropic in these lawsuits? WilmerHale. The same WilmerHale that Trump tried to destroy last year with an executive order stripping their government contracts and security clearances — because they kept Robert Mueller on their payroll after his investigation into Trump. That order was ruled unconstitutional in May 2025. The New Republic has the full story on how the same firm Trump tried to silence is now leading Anthropic’s legal charge.
On the scientific side, more than a dozen researchers from OpenAI and Google DeepMind — Anthropic’s two biggest competitors — filed an amicus brief in their personal capacities supporting Anthropic. Their argument: the supply chain risk designation could harm U.S. competitiveness in AI and chill critical public discourse about the technology’s risks.
Even rivals are stepping in. That tells you something about how the broader AI community reads this situation.
Why This Matters Far Beyond Anthropic
Whatever you think about Anthropic, Claude, or the Trump administration’s AI policy, this case has implications that extend far beyond one company’s government contracts.
If the Pentagon can designate any American tech company a national security risk as retaliation for policy disagreements, no AI developer is safe. The next target could be any company that refuses to remove safety restrictions. Or adds them. Or simply says something a government official dislikes on social media.
Several analysts and executives have pointed out that weakening Anthropic hands a significant advantage to China in the global AI race. Anthropic’s own lawsuit makes this argument directly: the government is “seeking to destroy the economic value created by one of the world’s fastest-growing private companies” in a field that is of vital national and global significance.
Complicating everything further: the Pentagon is still using Claude. CBS News reported that Claude remains in use during ongoing military operations. The six-month phaseout Hegseth announced is still in effect, meaning the government is simultaneously blacklisting and depending on the same technology — a contradiction that legal experts at Just Security say “undermines his assertion of any true supply chain risk.”
“The consequences of this case are enormous. The federal government retaliated against a leading frontier AI developer for adhering to its protected viewpoint on a subject of great public significance — AI safety and the limitations of its technology.” — Anthropic lawsuit
What Happens Next
Courts move slowly, but Anthropic is asking for injunctive relief — meaning a judge could block the Pentagon’s enforcement of the designation while the case proceeds. According to TechCrunch, Anthropic is also seeking a temporary restraining order to halt enforcement immediately.
The Pentagon declined to comment on the litigation, as is standard practice. The White House issued a statement saying the president “will never allow a radical left, woke company to jeopardize our national security.”
Anthropic, for its part, said it remains committed to supporting national security work and described the lawsuits as a last resort, not an end to negotiation. “We will continue to pursue every path toward resolution, including dialogue with the government,” the company said in a statement.
Microsoft and Google have both said they can continue non-defense-related work with Anthropic. But defense contractors like Lockheed are already gone, and that business may not come back easily regardless of how the courts rule.
The irony at the center of all this is hard to ignore. The company that was, until recently, the only AI lab cleared to operate on classified U.S. government networks — chosen precisely because of its safety focus — is now officially a threat to national security. Because it still believes in safety.
The Bottom Line for AI Marketers and Businesses
For those of us who work in the AI space, this case is a live demonstration of something we talk about in the abstract all the time: AI governance matters, and the rules of the road are still being written.
How this case resolves will shape what AI companies can and cannot commit to in their terms of service. It will influence what enterprise buyers can expect from their AI vendors in high-stakes deployments. And it will determine whether AI safety commitments are worth anything in practice, or whether they evaporate the moment a powerful enough customer says otherwise.
Watch this one closely. It will not stay in the courtroom.
Frequently Asked Questions
What is Anthropic’s lawsuit against the Pentagon about?
Anthropic filed two federal lawsuits on March 9, 2026 against the Department of Defense and other federal agencies after the Pentagon designated the company a “supply chain risk to national security.” The designation followed Anthropic’s refusal to allow its AI model Claude to be used for autonomous weapons or mass surveillance of U.S. citizens without restrictions. Anthropic argues the designation is unconstitutional retaliation for protected speech and exceeds the Pentagon’s statutory authority.
Why did the Pentagon label Anthropic a supply chain risk?
Defense Secretary Pete Hegseth issued the designation on March 5, 2026 after contract negotiations broke down. The Pentagon wanted Anthropic to agree to use Claude for “all lawful purposes” in military operations. Anthropic refused, citing its position that Claude cannot currently be safely and reliably deployed for autonomous lethal weapons systems or mass domestic surveillance. The supply chain risk label, historically applied only to foreign-linked companies, was applied to Anthropic for the first time in its history being used against a U.S.-based firm.
What could Anthropic’s lawsuit mean for AI companies?
If Anthropic loses, it could signal that AI companies cannot enforce safety restrictions in government contracts without risking financial ruin or blacklisting. If Anthropic wins, it would establish legal precedent that the government cannot use procurement power to punish companies for their stated AI safety policies. Either outcome will significantly shape how AI developers approach government contracts, safety commitments, and public advocacy going forward.
How does the Anthropic-Pentagon dispute affect OpenAI?
OpenAI signed a deal with the Pentagon hours after Anthropic’s negotiations collapsed, agreeing to provide its models for military use. The deal drew widespread criticism for appearing opportunistic, and OpenAI later acknowledged the announcement looked “sloppy” and said it was renegotiating terms. Anthropic CEO Dario Amodei has since called OpenAI leadership’s behavior deceptive in internal communications, and relations between the two companies have significantly deteriorated.
