Trump Administration Rejects Anthropic’s Request for Pause on New Penalties Amid AI Startup Legal Fight
During AI startup Anthropic’s first court hearing challenging sanctions imposed by the Trump administration, the company’s core request was straightforward: secure a formal commitment from the federal government that it would not impose new additional penalties against the firm. That request was outright denied.
“I am not prepared to offer any commitments on that issue,” James Harlow, a U.S. Department of Justice attorney, told U.S. District Judge Rita Lin during Tuesday’s videoconference hearing.
In fact, the federal government is already preparing another move designed to fully cut Anthropic off from doing business with federal agencies. According to a White House source familiar with the unannounced plan (who spoke on condition of anonymity, as they were not authorized to discuss the policy publicly), President Trump is currently finalizing an executive order that would formally ban all federal use of Anthropic’s AI tools. Axios was the first outlet to report details of the draft order.
Tuesday’s hearing was tied to one of two federal lawsuits Anthropic filed against the Trump administration last Monday. In the suits, the startup argues the government unconstitutionally labeled it a national supply chain threat, effectively turning it into a pariah across the tech industry. Anthropic says billions of dollars in current and projected revenue are already at risk, as existing and prospective clients have pulled out of deals and demanded renegotiated terms for existing agreements.
Anthropic is asking the court to issue a preliminary injunction that would suspend the government’s risk designation and block the administration from implementing any further punitive actions against the company.
Tuesday’s court session was held exclusively to set a timeline for the preliminary injunction hearing, and Anthropic has pushed aggressively for an early hearing to stop growing damage to its business. Michael Mongan, an Anthropic attorney at the firm WilmerHale, told Lin his team would accept a delay to April only if the Trump administration agreed to hold off on new punitive steps in the interim. “The actions of defendants are causing irreparable injuries, and those injuries are mounting day by day,” Mongan said.
After Harlow declined the requested commitment, Lin moved the hearing up to March 24 in San Francisco. While the date is earlier than the government proposed, it is still later than Anthropic had requested. “The case is quite consequential from both sides, and I want to make sure I’m deciding on an expedited record but also a full record,” the judge explained.
Scheduling for Anthropic’s second lawsuit, filed in Washington, D.C., remains on pause while the company pursues an administrative appeal through the Department of Defense, a process widely expected to end in rejection this Wednesday.
The months-long dispute between the Pentagon and Anthropic began when the AI startup refused to sign a required clause confirming its existing technology could be used by the military for any lawful purpose. The firm says it objected out of fear its technology would be used for mass surveillance of U.S. citizens and to launch missile strikes without direct human oversight. The Pentagon has countered that all decisions regarding military use of technology fall exclusively under its authority.
Multiple legal experts specializing in federal contracts and U.S. constitutional law argue the administration’s crackdown on Anthropic fits a broader pattern of misusing legal authority to punish perceived political opponents. Past targets of this pattern, they say, have included universities, media outlets, and even law firms—including WilmerHale, which is representing Anthropic in this case. Most experts believe Anthropic has a strong legal argument to prevail, but the biggest hurdle will be overcoming the deference U.S. courts typically grant to executive branch national security claims, particularly amid ongoing global conflicts.
“If this is a one-off, you might give the president some deference,” said Harold Hongju Koh, a Yale Law School professor and former official in the Barack Obama administration who has published analysis on the Anthropic case. “But now, it’s just unmistakable that this is just the latest in a chain of events related to a punitive presidency.”
David Super, a constitutional law professor at Georgetown University Law Center, notes that the legal provisions the Pentagon used to sanction Anthropic were originally drafted to protect the U.S. from sabotage by foreign enemies. “It is an absurd stretch of the English language to equate ‘does not agree to every demand of Pete Hegseth’ with ‘sabotage,’” Super said, referencing the current Secretary of Defense. “The U.S. Supreme Court has repeatedly warned against this sort of repurposing [of statutory authority], both in the recent case striking down the president's tariffs and in earlier cases striking down President Biden's actions, such as student loan forgiveness or the pandemic-era eviction moratorium.”
While the lawsuits work their way through the court system, the broader tech industry is already navigating a set of thorny practical dilemmas. Software companies that build products on top of Anthropic’s Claude AI suite are unsure if they need to seek out alternative tools. Competitors OpenAI and Google have already moved forward to secure Pentagon contracts to replace Anthropic, despite internal pressure from their own employees to push back against the government’s terms for AI use.
Zohra Tejani, a partner at law firm Seyfarth Shaw who advises tech companies on federal contracts, says Anthropic may ultimately succeed in having the supply chain risk label thrown out and resume normal business with most of its private clients. Even so, she notes, the company is unlikely to win back federal contracts under the current administration.
The government’s multi-front attack on Anthropic and its ethical AI principles could leave other government contractors too afraid to push back against administration demands—even if courts ultimately rule the government’s actions unlawful. Many observers see that outcome as the Trump administration’s ultimate goal.
“The Pentagon is sending a message to every other AI company: If you defy the Pentagon, you risk nationalization and heavy-handed government intervention,” said Christoph Mlinarchik, a former Pentagon contracting officer who now advises federal suppliers. “The Pentagon does not want to cede veto or moral authority to contractors, no matter the flavor of technology.”
Makena Kelly contributed reporting.
