Now loading...
U.S. Secretary of War Pete Hegseth announced on social media platform X on Thursday that he’s instructing the Department of War to label AI developer Anthropic as a potential risk to the nation’s supply chain. The move comes after extended talks between the two sides broke down, primarily over Anthropic’s insistence on restrictions against using its Claude AI model for widespread surveillance of U.S. citizens at home and for weapons that operate without human oversight.
Anthropic stated that it has not been formally notified by the Department of War or the White House about the outcome of the discussions. The company emphasized its efforts to negotiate fairly, noting its willingness to allow all other legal applications of its technology for defense purposes. According to Anthropic, these specific limitations have not hindered any official operations so far.
The firm explained its stance by arguing that existing advanced AI systems lack the dependability needed for independent weaponry, which could put soldiers and non-combatants at risk. It also views large-scale monitoring of the domestic population as an infringement on core civil liberties.
Such a risk classification would mark a historic step, one typically aimed at foreign rivals rather than domestic businesses. Anthropic expressed regret over the situation, highlighting its role as the initial leading AI provider to integrate systems into secure government environments starting in June 2024 and its commitment to ongoing assistance for U.S. military personnel.
The company described the potential designation as flawed both in law and as a model for future dealings between private firms and federal agencies. It vowed to resist any such label through legal action and affirmed that pressure from the Department of War will not alter its policies on surveillance or self-operating arms.
Regarding impacts on users, Anthropic clarified that Secretary Hegseth’s suggestion of broader restrictions on partners lacks legal backing. Under the relevant statute, 10 U.S.C. 3252, any prohibition would apply solely to Claude’s integration in Department of War projects, leaving other applications untouched. This ensures no changes for personal users or those with standard commercial agreements, while Department of War affiliates would face limits only on official contract-related tasks.
Anthropic’s customer service and outreach staff are available to address concerns. The company thanked supporters from the tech sector, government officials, former service members, and the wider community, underscoring its focus on minimizing disruptions for clients and facilitating an orderly handover for defense activities to safeguard troops and national efforts.
