A federal judge has signaled deep skepticism over the Department of War's decision to label Anthropic a supply-chain risk, suggesting the move may constitute illegal retaliation. During a Tuesday hearing in San Francisco, US District Judge Rita Lin described the government's actions as an apparent attempt to cripple the AI developer after it sought limitations on military deployments of its models.

Anthropic argues the Trump administration weaponized national security designations because the company pushed back on how its technology could be used in combat scenarios. The labeling threatens to scare off commercial clients, prompting Anthropic to seek an immediate injunction. Lin noted that while Defense Secretary Pete Hegseth decides vendor suitability, expanding restrictions beyond government contracts raises legal red flags. Hegseth previously ordered all military contractors to cease commercial work with Anthropic, a mandate his own attorneys admitted lacked legal backing during the hearing.

The government claims Anthropic could manipulate its software if disputes arise, posing a threat to operational reliability. However, Lin questioned why less punitive measures weren't explored before invoking authorities typically reserved for hostile foreign actors. Meanwhile, the Department of War is transitioning to alternatives from Google, OpenAI, and xAI.

For machine learning engineers, this case underscores the friction between safety protocols and government deployment demands. It raises questions about model ownership once integrated into defense infrastructure. Lin's ruling on the temporary pause is expected within days, potentially setting a precedent for leverage AI labs retain over deployed systems. The broader legal challenge continues in Washington, DC.

Source: Wired