The U.S. Department of Defense has designated Anthropic's Claude artificial intelligence model as a supply chain risk, citing embedded policy preferences that could compromise defense procurement. This unprecedented action requires defense contractors to certify they do not use Claude in Pentagon-related work.
Pentagon's Stance
Emil Michael, the Pentagon's Chief Technology Officer, explained on CNBC's "Squawk Box" that Claude's inherent policy differences could "pollute" the supply chain, potentially leading to ineffective weapons and equipment. He emphasized that the risk arises from the model's constitution and policy preferences being fundamentally "baked in."
- Key points from Michael:
- The designation aims to ensure warfighters receive reliable protection and avoid compromised products.
- This marks the first time an American company has been labeled a supply chain risk, a measure historically reserved for foreign adversaries.
