A federal appeals court has refused to halt the Trump administration's blacklisting of Anthropic, escalating a high-stakes confrontation between AI ethics and government power.
Three Republican-appointed judges on the US Court of Appeals for the District of Columbia Circuit denied Anthropic's emergency motion for a stay on Wednesday, allowing the Trump administration's punitive measures against the AI company to remain in effect. The panel included two Trump appointees, Gregory Katsas and Neomi Rao, both of whom held senior roles in Trump's first administration. Katsas served as deputy counsel to the president, while Rao worked in the Office of Management and Budget.
The court did grant Anthropic's request to expedite the case, scheduling oral arguments for May 19. But the denial of the stay means the company continues to face severe federal restrictions in the interim, restrictions that could materially damage its business relationships and revenue pipeline.
At the center of this dispute is Anthropic's refusal to allow its Claude AI models to be deployed for autonomous warfare systems and mass surveillance of American citizens. The company has framed this as a First Amendment issue, arguing that it cannot be punished by the government for exercising its right to choose how its technology is used. The Trump administration, meanwhile, has characterized Anthropic as a national security liability.
President Trump directed all federal agencies to cease using Anthropic technology. Defense Secretary Pete Hegseth went further, officially labeling the company a supply-chain risk to national security, a designation that effectively bars military contractors from maintaining business relationships with Anthropic. For a company competing in a market where government and defense contracts represent a significant and growing revenue stream, the financial implications are substantial.
The blacklist creates a cascading problem for Anthropic that reaches well beyond lost federal deals. When a major defense contractor like Lockheed Martin or Raytheon is told it cannot work with a specific vendor, other large enterprises take notice. The supply-chain risk designation functions as a warning label, one that could spook potential private-sector clients who worry about regulatory scrutiny or political blowback. As Ars Technica first reported, the ruling represents a significant setback for Anthropic even though it is only one of two legal challenges the company has filed against the administration.
Anthropic has had more success in its parallel case, though the details of that proceeding remain less visible. What is clear is that the company is fighting a two-front legal war, burning through cash and executive attention that would otherwise go toward product development and market expansion. For a startup valued at roughly $18 billion following its Amazon-backed funding round, sustained legal battles are more than a distraction. They are an existential tax on growth.
A Defining Moment for AI Governance
This case cuts to the heart of a question the technology sector has been dodging for years: who decides how AI is used when the government is the customer? Anthropic built its brand on safety-first AI development. Its founding team, led by former OpenAI vice president of research Dario Amodei, positioned the company as the responsible alternative in a market racing toward capability without guardrails. Refusing to sell Claude for autonomous warfare or domestic surveillance is entirely consistent with that positioning. But consistency does not protect you from a president willing to use executive power to punish dissent.
The larger concern for the AI industry is precedent. If the courts ultimately uphold the administration's authority to blacklist companies based on their refusal to participate in specific government applications, every AI startup in the United States faces a choice between complicity and exclusion. Companies like OpenAI, Google DeepMind, and Cohere are all navigating similar pressure from the Pentagon and intelligence agencies eager to integrate AI into defense operations. Anthropic's fight could determine how much leverage private companies actually have when they say no.
The May 19 oral arguments will offer the first real window into how the appellate court views the constitutional questions at play. Until then, Anthropic operates under a cloud that no amount of technical brilliance can clear. For founders and investors watching from the sidelines, the lesson is already taking shape. In the current political environment, your terms of service are only as strong as your willingness to defend them in court.
