Anthropic Sues Pentagon Over Supply Chain Risk Label

Anthropic said it will challenge in court a U.S. Department of Defense designation that labels the company a “supply chain risk,” setting up a high-stakes legal fight between a leading artificial intelligence developer and the Pentagon.

The dispute centers on a formal notice from the Department of Defense that Anthropic has been officially told it is considered a supply chain risk. Anthropic, the maker of the Claude AI model, has said it plans to contest that designation in court. Public reports describe the decision as a Pentagon move that could restrict the company’s access to Defense-related work and tighten limits on use of its technology.

The development comes amid mounting attention on how federal agencies evaluate and manage risks tied to AI systems, their vendors, and the software supply chains that support them. A supply chain risk label can carry sweeping consequences in government procurement, affecting contracting eligibility, partner relationships, and ongoing or future deployments.

Anthropic’s situation has also drawn scrutiny because Claude has been used in Iran, according to related coverage, even as the Pentagon has flagged Anthropic itself as a supply chain concern. That juxtaposition has intensified questions around how AI tools are adopted, monitored, and controlled across borders, and how national security agencies reconcile real-world usage with internal risk assessments.

The case matters beyond one company because it could shape how the U.S. government sets standards for AI vendors and how companies can contest security designations that affect their business. In the AI sector, where commercial models are increasingly integrated into enterprise and government workflows, a federal risk label can function like an effective ban even without a criminal allegation or a formal enforcement action.

It also reflects a broader tension between Silicon Valley and Washington as agencies accelerate AI adoption while simultaneously building stricter guardrails. The Pentagon’s assessments often rely on sensitive information and internal reviews that are not fully public, while companies argue they need clearer processes and the ability to respond to claims that can cut them off from major customers.

Next, Anthropic is expected to bring its challenge in court, seeking to overturn or narrow the Defense Department’s designation. The legal process will determine what evidence can be presented, what information may remain classified, and whether the Pentagon followed required procedures in reaching its supply chain risk conclusion.

The Defense Department, for its part, will likely defend its authority to assess vendor risk and to restrict access to systems and contracts it deems sensitive. Any court-ordered changes could affect not only Anthropic’s status but also how similar risk determinations are made and reviewed for other technology providers.

As the dispute moves toward litigation, it will test how transparent the government must be when it labels an AI company a security risk—and how much power that label carries in shaping who gets to build the tools used inside the U.S. national security apparatus.