Pentagon AI Dispute: Anthropic Refuses to Back Down
▼ Summary
– The U.S. military has given Anthropic a deadline to grant unrestricted access to its AI model or face being declared a supply chain risk or having the Defense Production Act (DPA) invoked against it.
– Anthropic is refusing to compromise on its policies against using its technology for mass surveillance or fully autonomous weapons.
– Invoking the DPA in this AI dispute would mark a significant expansion of the law and could be seen as an attack on stable commerce, according to a policy expert.
– The Pentagon lacks a backup AI provider for classified systems, making it dependent on Anthropic and explaining its aggressive stance.
– The conflict is set against ideological friction, with some administration officials criticizing Anthropic’s safety policies as “woke.”
A high-stakes confrontation is escalating between the Pentagon and leading artificial intelligence company Anthropic, with the Defense Department threatening to use extraordinary legal powers to compel access to the firm’s advanced AI models. Officials have set a Friday deadline for Anthropic to grant unrestricted military access or face being labeled a national security risk, a move that could cripple the company’s operations and set a major precedent for government intervention in the tech sector.
Defense Secretary Pete Hegseth delivered this ultimatum directly to Anthropic CEO Dario Amodei, presenting two stark options. The Pentagon could formally designate Anthropic a “supply chain risk,” a tag typically applied to foreign adversaries, or it could invoke the Defense Production Act (DPA). This law grants the president sweeping authority to mandate that private companies prioritize national defense needs, a tool last used extensively during the COVID-19 pandemic to ramp up production of medical equipment.
At the heart of the standoff are Anthropic’s self-imposed ethical guardrails. The company has consistently refused to allow its technology to be used for mass surveillance of American citizens or for developing fully autonomous weapon systems. Pentagon leaders counter that military applications of technology should be governed solely by U.S. law and the Constitution, not by the internal policies of private contractors.
Invoking the DPA to override a company’s AI safety policies would represent a dramatic and controversial expansion of the law. According to Dean Ball, a senior fellow at the Foundation for American Innovation and a former White House AI policy advisor, such an action signals dangerous instability. “It would basically be the government saying, ‘If you disagree with us politically, we’re going to try to put you out of business,’” Ball stated. He warned that this approach could severely damage the United States’ reputation as a stable hub for global commerce and innovation.
The conflict is fueled by ideological divisions within the administration, where some officials, including AI czar David Sacks, have publicly dismissed Anthropic’s safety commitments as overly restrictive or “woke.” Despite the intense pressure, Anthropic shows no signs of backing down, with reports indicating the company will not relax its usage restrictions.
The Pentagon’s aggressive stance may be driven by a critical vulnerability. Anthropic is currently the only frontier AI lab with the necessary clearances for the Department of Defense’s classified systems, creating a precarious single-source dependency. While the Pentagon has reportedly secured a deal to use xAI’s Grok model in some classified contexts, it lacks a ready-made alternative to Anthropic’s full capabilities.
This lack of a backup plan puts the Defense Department in a bind. Ball emphasized the severity of the situation, noting, “If Anthropic canceled the contract tomorrow, it would be a serious problem for the DOD.” He pointed out that this reliance appears to contradict a recent National Security Memorandum advising federal agencies against depending on a single provider for classified AI systems. The outcome of this showdown will have profound implications for the future of AI governance, military procurement, and the balance of power between the state and the private technology sector.
(Source: TechCrunch)
