Pentagon Labels Anthropic as Supply Chain Risk

Pentagon Notifies Anthropic of Supply Chain Risk

On March 4, 2026, Anthropic received a letter from the United States Department of Defense (Pentagon) indicating that the company has been identified as a potential risk in the supply chain for defense-related contracts. This was publicly confirmed by CEO Dario Amodei in an official statement released a day later.

According to Amodei, the company disagrees with the legal interpretation used by the government and intends to challenge the measure in court. Nevertheless, he asserts that the decision is limited in scope and does not prevent clients from using Claude, the company's AI model, outside of contracts directly tied to the Department of Defense.

Measure Only Affects Specific Military Contracts

Amodei explained that the government's letter cites 10 USC 3252, which allows the Department of Defense to identify suppliers deemed risks in the supply chain. However, this law was created to protect government systems, not to punish companies, meaning the restrictions should be as non-invasive as possible.

In practice, the designation would only have an impact when Claude is directly integrated into projects linked to Pentagon contracts. Other uses of the technology by private companies or organizations also working with the government would not be affected.

The CEO stated that the language of the government's letter itself confirms this limitation. Thus, he believes that most of Anthropic's clients should not face any immediate impact.

Company Continues Collaboration with US Military

Despite the impasse, Amodei mentioned that discussions between Anthropic and the Department of Defense are ongoing. In recent days, representatives from both parties have explored alternatives to maintain technical support and ensure a potential transition if the restrictions remain.

The company claims to have worked with the US government on various AI applications focused on national security, including:

• Intelligence analysis
• Military modeling and simulation
• Operational planning
• Cyber operations

According to Amodei, these projects directly assist military forces and national security experts in real operations.

Debate Over Limits on Military Use of AI

In the statement, the CEO emphasized that Anthropic does not intend to participate in military operational decisions. The company maintains two main restrictions for the use of its models: fully autonomous weapons systems and mass domestic surveillance.

These limitations are part of the company's public policy on the responsible use of artificial intelligence. Amodei states that they pertain to the strategic level of technology application, not the direct operation of armed forces.

Context Includes Recent Tensions in Military AI Sector

The statement also mentions an incident that took place days earlier, when an internal company publication was leaked to the press following announcements involving military AI contracts in the United States. Among them was an agreement announced between the Pentagon and OpenAI, which was later described by the company itself as having been communicated confusingly at first.

While the legal dispute is unresolved, Amodei stated that Anthropic will continue to make its models available to the Department of Defense and the national security community for a symbolic cost, in addition to maintaining technical support during the transition period.