ACT has filed amicus briefs in District Courts in California and D.C. in support of Anthropic’s lawsuit against the U.S. Department of War. In response to Anthropic refusing to allow its products to be used for autonomous lethal action or the mass surveillance of Americans, Secretary Hegseth in March 2026 attempted to declare Anthropic’s Claude AI models a supply chain risk. A supply chain risk designation would ban contractors (and their subcontractors), working with the military and associated agencies, from using Claude AI or even conducting “any commercial activity” with Anthropic. In their lawsuit, Anthropic focused on the major First Amendment concerns raised by the Department’s action, and those arguments are certainly important.

ACT weighed in to explain to the courts the significant practical problems with implementing a supply chain risk decree against Claude and the disproportionate negative effect it would have on startups and small technology businesses. Historically, this type of majorly disruptive designation goes through a public comment and rulemaking process. During that process, the government would provide significant details on their rationale, collect public feedback from interested parties, and respond to concerns raised. The supply chain risk designation has gone through none of the statutorily required process. If the supply chain risk designation isn’t struck down, ACT members are going to be on the front lines of attempting to comply with this nebulous order.

Unlike a piece of communications equipment or even an entire company’s output of products and services, there are enormous difficulties with trying to ban the use of a specific large language model. Here are a few examples from our discussions with ACT members that formed the backbone of our brief:

  • A two-person startup selling logistics software to a defense prime contractor uses Claude Code to write their entire test suite; nothing in the final product they ship identifies the use of an AI tool, as it is functionally indistinguishable from hand-written code. Is that startup conducting “any commercial activity” with Anthropic? What about customers who buy this product?
  • A defense contractor’s developer pulls in more than 800 open-source dependencies. How are they supposed to know which of those packages were refactored, debugged, or documented using Claude Code by their maintainers?
  • A mid-tier supplier to a defense prime contractor uses a software-as-a-service (SaaS) vendor’s project management tool, and that SaaS vendor’s engineering team rebuilt their interface using Claude Code last quarter. That supplier has no knowledge, and no contractual visibility into, how their vendor’s software was built. Are they complying with the supply chain risk order? How can they know either way?
  • A government contractor uses a popular open-source database library and, and six months ago, the home hobbyist who works on the project used Claude Code to optimize a query parser. Is every downstream contractor now implicated?
  • A developer building medical records software for a hospital system, which also has a contract with the Department of Veterans Affairs, uses Claude frequently to write Health Insurance Portability and Accountability Act (HIPAA)-compliant data handlers. Does this activity put the hospital system in legal jeopardy?

Even with lengthy deliberation and information exchange between policymakers and affected stakeholders, none of these scenarios have easy answers. When the government hastily orders a ban with no process at all, the consequences could be dire, particularly for small business developers. If supply chain risk designations like this one are allowed to become a commonly used tool and today’s premier productivity technology can become illegal to use tomorrow, developers will be left with no choice but to hedge against or even reject new products that could increase efficiency and lower cost in favor of outdated tools less likely to draw the wrong kind of government attention.

The District Court in California agreed. On March 26, Judge Rita Lin granted Anthropic’s request for a preliminary injunction on the Department’s supply chain risk designation, repeatedly citing ACT’s brief in justifying her ruling.

The District Court in D.C., however, did not grant a stay, ordering on April 8 that the parties instead proceed to briefing on the merits on an expedited schedule, with Anthropic’s and supporters’ briefs due April 22.  At the same time, the Department of War appealed Judge Lin’s injunction to the Ninth Circuit Court of Appeals, which has set a briefing schedule from early to late May. ACT is planning to file briefs again at both opportunities.

As this groundbreaking case moves forward on both the east and west coasts, we will continue to work to make sure ACT members’ voices are heard.

Read ACT’s full amicus brief here