Anthropic Responds After US Military Designates It as a ‘Supply Chain Threat’

On Friday, United States Secretary of Defense Pete Hegseth ordered the Pentagon to classify Anthropic as a “supply-chain risk,” causing upheaval in Silicon Valley and prompting numerous companies to reassess their use of one of the industry’s leading AI models.

“Effective immediately, no contractor, supplier, or partner engaged with the United States military may undertake any commercial activities with Anthropic,” Hegseth announced via social media.

This designation followed weeks of strained discussions between the Pentagon and Anthropic regarding the military’s potential uses of the startup’s AI models. In a blog post earlier this week, Anthropic contended that its agreements with the Pentagon should prohibit its technology from being utilized for mass domestic surveillance or fully autonomous weapons. Conversely, the Pentagon sought Anthropic’s consent to allow military applications of its AI for “all lawful uses” without specifying exceptions.

Designating a company as a supply chain risk enables the Pentagon to limit or exclude specific vendors from defense contracts believed to present security threats, particularly regarding foreign ownership, control, or influence. This measure aims to safeguard sensitive military systems and information from potential breaches.

In a subsequent blog post on Friday evening, Anthropic stated that it would “contest any supply chain risk designation in court,” arguing that such a classification would “set a precarious precedent for any American company negotiating with the government.”

The company also indicated it had not received direct communication from the Department of Defense or the White House concerning the usage of its AI models.

“Secretary Hegseth suggests this designation would inhibit anyone conducting business with the military from engaging with Anthropic. However, the Secretary lacks the statutory authority to support this claim,” the company asserted.

The Pentagon refrained from making any comments.

“This is the most shocking, damaging, and overreaching action I have ever witnessed from the United States government,” stated Dean Ball, a senior fellow at the Foundation for American Innovation and former senior policy advisor for AI at the White House. “We have effectively sanctioned an American company. If you’re an American, you should consider whether you want to remain here in 10 years.”

People throughout Silicon Valley took to social media, expressing similar sentiments of shock and regret. “The individuals managing this administration are impulsive and vindictive. This behavior is self-explanatory,” stated Paul Graham, founder of the startup accelerator Y Combinator.

Boaz Barak, an OpenAI researcher, remarked in a post that “crippling one of our leading AI firms is a major misstep. I sincerely hope reason prevails and this announcement is rescinded.”

In contrast, OpenAI CEO Sam Altman revealed on Friday night that the company had reached an agreement with the Department of Defense to utilize its AI models in classified environments, seemingly with specific exceptions. “Two of our fundamental safety principles prohibit domestic mass surveillance and ensure human accountability in the use of force, including autonomous weapons systems,” Altman stated. “The DoD concurs with these principles, incorporates them into law and policy, and we have reflected them in our agreement.”

Confused Customers

In its Friday blog post, Anthropic clarified that a supply chain risk designation, as per the authority 10 USC 3252, applies solely to Department of Defense contracts directly involving suppliers and does not extend to how contractors utilize its Claude AI software for other clients.

Three experts in federal contracts indicated that it is currently impossible to ascertain which Anthropic customers, if any, are now required to sever ties with the company. Hegseth’s announcement “is not anchored in any discernible law at this moment,” noted Alex Major, a partner at the law firm McCarter & English, which collaborates with tech firms.