Defense Department and Anthropic Clash Over A.I. Safety

The Defense Department and the A.I. company Anthropic have been negotiating a contract for use of the company’s systems on classified Pentagon networks, and discussions this week turned into a public dispute. The Pentagon signaled it might declare Anthropic a supply-chain risk, a move that would sever ties, and the company scrambled to determine what had prompted the escalation.

At the center of the disagreement are limits Anthropic sought on how its technology is deployed. The company told defense officials it did not want its A.I. used for domestic mass surveillance or in fully autonomous weapons, and its chief executive has called mass surveillance and mass propaganda “entirely illegitimate.” Pentagon officials pushed back, objecting to restrictions on how the military would use the tools and accusing Anthropic of catering to its workforce’s preferences.

Anthropic’s systems have been part of a $200 million Defense Department A.I. pilot program for over a year.

United States

anthropic, pentagon, defense department, ai safety, supply chain, classified networks, mass surveillance, autonomous weapons, mass propaganda, pilot program