Pentagon contract dispute creates uncertainty for AI startups

Negotiations between the Pentagon and artificial-intelligence startup Anthropic over the military’s use of its Claude technology have collapsed. The breakdown in talks was compounded last week when the Trump administration designated Anthropic a supply-chain risk—a formal government designation regarding a vendor. Anthropic said it would fight the supply-chain risk designation in court. The dispute arose because the Pentagon, which serves as the headquarters for the US Department of Defense, sought to change existing terms on an existing contract. Although Anthropic’s technology is still being used by the military and is considered a crucial technology, the sudden attempt to alter established contract terms has introduced deep regulatory uncertainty for private tech vendors.

Following the collapse of the Anthropic negotiations last week, OpenAI quickly announced its own deal with the Department of Defense. The announcement triggered immediate internal and external backlash. At least one OpenAI executive resigned over concerns that the deal was rushed without appropriate guardrails in place. Meanwhile, consumer backlash was swift: ChatGPT uninstalls surged 295% after OpenAI locked in the deal with the Department of Defense. This consumer migration temporarily pushed Anthropic’s Claude to the top of the App Store charts, illustrating how public sentiment can shift rapidly when consumer brands venture into military contracting.

The contrasting outcomes for these high-profile AI firms highlight the unique scrutiny faced by startups entering the defense sector. While established contractors like General Motors have long manufactured defense vehicles for the Army under the radar, consumer-facing AI companies face intense public debate. As TechCrunch podcast host Sean O’Kane noted, the debate centers on “how their technologies are being used or not being used to kill people.” This scrutiny complicates the standard dual-use model—defined as technology that has both civilian and military applications. Other dual-use contractors, such as Applied Intuition, are unlikely to back off from federal defense work because they do not face the same public spotlight.

Both OpenAI and Anthropic have publicly stated they want restrictions on how their AI is used, but Anthropic has taken a firmer stance against allowing the government to alter contract terms. This friction has led podcast host Kirsten Korosec to question whether startups might begin to hesitate before pursuing federal dollars. For consumer AI brands, the friction is further intensified by personal dynamics: the CEO of Anthropic and Emil Michael, the chief technology officer for the Department of Defense, reportedly do not like each other.

Why it matters

The Pentagon’s attempt to change existing contract terms, combined with the supply-chain risk designation, creates significant uncertainty for startups considering federal defense work. This situation highlights the unique scrutiny faced by high-profile AI companies contracting with the federal government.