In lawsuits filed on March 9, 2026, AI company Anthropic argues that neither President Donald Trump nor Defense Secretary Pete Hegseth has the authority to label it a “supply-chain risk” and effectively blacklist its products.
Anthropic filed a pair of lawsuits against the federal government for First Amendment violations. While Anthropic CEO Dario Amodei insists negotiations are still ongoing between his company and the Trump administration (a claim denied by Pentagon CTO Emil Michael), he also argued that the lawsuits were necessary to address violations of the guidelines set forth in 10 U.S.C. § 3252.
The backlash from the government followed negotiations for a $200 million contract that broke down in February 2026 over concerns that the DOD wanted to use Anthropic’s AI chatbot, Claude, for military operational decision-making. The suits claim that the narrow scope of 10 U.S.C. § 3252 applies only to firms aiding foreign adversaries and doesn’t apply to American companies. That hasn’t stopped Pentagon officials from canceling Anthropic’s contracts and informing other defense contractors that doing business with Anthropic could jeopardize their Pentagon work.
Guardrails and the Pentagon’s New AI Fight
Anthropic’s Claude is the company’s premier AI model. Anthropic’s technology and AI systems were among the first “frontier” large language models (LLMs) used in the nation’s classified systems and developed into custom models for other defense and national security contractors. Use of Claude became an integral part of intelligence analysis, operational planning, modeling and simulation, and cyber operations at the DOD.
Negotiations over a new $200 million contract between the DOD and Anthropic may have been considered a formality when they began, but in February 2026, things went awry. The artificial intelligence company refused to budge from guardrails against the use of Claude by the U.S. military for either mass surveillance of Americans or fully autonomous weapons, a pair of red lines they refused to cross.
Secretary Hegseth, who has argued that private companies don’t get to tell the government how to set policy, reportedly threatened to invoke the Defense Production Act (DPA) to compel Anthropic to provide the DOD a version of Claude without safeguards against certain types of military use. The DPA allows the president, in the interest of national defense, to direct privately owned companies. A recent example of this was when President Joseph Biden and President Trump, in separate actions, ramped up the supply of essential medical goods during the coronavirus pandemic.
Instead of applying the DPA, the administration abruptly changed course, declaring that Anthropic was now a “supply-chain risk” threatening national security. This essentially blacklisted Anthropic’s products, requiring federal agencies to remove the company’s AI tools and telling other defense contractors they needed to rid themselves of Claude as well. This left Amodei and Anthropic with little choice but to file suit.
Make Sure Not To Use AI To Write Any Legal Briefs, Okay?
The two lawsuits filed by Anthropic essentially make the same argument, with a shorter version filed in the D.C. Court of Appeals due to procedural requirements regarding one of the statutes invoked by the government. The suits acknowledge that the administration can choose whether to do business with Anthropic, but contend that it violated the company’s free speech protections under the First Amendment by labeling it a “supply-chain risk” over statements about its stance on AI safeguards.
Claiming that Congress, not President Trump or the DOD, has procurement powers, Anthropic charges that the government’s actions are less an accurate reflection of its powers under 10 USC 3252 and more acts of retaliation. The suit further argues that the company has demonstrated that its AI tools can be used by other defense contractors without posing a national security threat.
Also citing due process and Article II violations arising from President Trump’s ultra vires edicts, Anthropic is seeking, in addition to having Secretary Hegseth’s order stayed and vacated, to have the government enjoined from enforcing the blacklisting of its AI products.
One place the AI company has found allies is among its chief competitors. Both OpenAI and Google have voiced support for Anthropic in its fight. The company may want to temper its enthusiasm, as OpenAI announced an agreement with the government soon after President Trump ordered federal agencies to stop using Claude.
Related Resources
- Lindell Case Indicates Artificial Intelligence and the Legal System Are Still a Work in Progress (FindLaw’s Practice of Law)
- Artificial Intelligence Company Pays $1.5 Billion in Settlement, but Did It Really Lose? (FindLaw’s Courtside)
- Reddit’s Lawsuit Over Data-Scraping Could Reshape the Future of AI (FindLaw’s Courtside)
