Tech Workers Demand DOD, Congress Drop Anthropic's Supply Risk Label.

A collective of hundreds of technology professionals has endorsed an open letter, advocating for the Department of Defense to revoke its classification of Anthropic as a "supply chain risk." The letter further implores Congress to intervene and "examine whether the use of these extraordinary authorities against an American technology company is appropriate."

Prominent technology and venture capital entities, such as OpenAI, Slack, IBM, Cursor, and Salesforce Ventures, are among the signatories. This collective action stems from a recent disagreement between the DOD and Anthropic, which arose last week when the AI laboratory declined to grant the military unfettered access to its artificial intelligence systems.

During its discussions with the Pentagon, Anthropic established two non-negotiable conditions: its technology should not be employed for widespread surveillance of American citizens, nor should it power autonomous weapon systems capable of making targeting and engagement decisions without human oversight. While the DOD stated it had no immediate intentions for such applications, it also asserted that its operations should not be constrained by a vendor's stipulations.

Following Anthropic CEO Dario Amodei's steadfast refusal to yield to Hegseth’s demands, President Donald Trump on Friday instructed federal agencies to cease utilizing Anthropic’s technology after a six-month transition phase. Hegseth affirmed his intent to follow through on his warnings by designating Anthropic as a supply chain risk—a classification typically applied to foreign adversaries, which would effectively bar the AI firm from engaging with any agency or company conducting business with the Pentagon.

In a statement posted on Friday, Hegseth declared: “Effective immediately, no contractor, supplier, or partner that does business with the United States military may conduct any commercial activity with Anthropic.”

However, a social media declaration alone does not automatically confer a "supply chain risk" status upon Anthropic. The government is required to conduct a formal risk assessment and inform Congress before military partners are compelled to sever ties with Anthropic or its offerings. In a blog post, Anthropic characterized the proposed designation as both “legally unsound” and stated its intention to “challenge any supply chain risk designation in court.”

Within the industry, the administration's actions toward Anthropic are widely perceived as unduly severe and a direct act of retaliation.

The open letter contends that "When two parties cannot agree on terms, the normal course is to part ways and work with a competitor." It continues, warning that "This situation sets a dangerous precedent. Punishing an American company for declining to accept changes to a contract sends a clear message to every technology company in America: accept whatever terms the government demands, or face retaliation.”

Beyond the immediate apprehension regarding the government's severe handling of Anthropic, a broader concern persists within the industry concerning potential governmental overreach and the deployment of AI for malicious objectives.

Boaz Barak, a researcher at OpenAI, articulated in a social media post on Monday that preventing governments from employing AI for mass surveillance constitutes his “personal red line” and should similarly be “all of ours.”

Shortly after Trump's public criticism of Anthropic, OpenAI disclosed that it had independently secured an agreement for its models to be deployed within the DOD’s classified environments. OpenAI CEO Sam Altman affirmed last week that his firm adheres to the same fundamental principles as Anthropic.

Barak concluded by stating, “If anything good can come out of the events of the last week, it would be if we in the AI industry start treating the issue of using AI for government abuse and surveilling its own people as a catastrophic risk of its own right.” He further urged, “We have done a good job of evaluations, mitigations, and processes, for risks such as bioweapons and cyber security. Let’s use similar processes here.”