Pentagon officials are at odds with Anthropic, the company behind an $200 million contract to develop AI capabilities for U.S. national security. Defense Secretary Pete Hegseth has given CEO Dario Amodei until Friday to grant the military full access to its artificial intelligence model, Claude.
The issue is over control of the AI technology, which the Pentagon wants to use in its operations without restrictions. Anthropic’s CEO had previously asked for guardrails that would prevent mass surveillance of Americans, but Defense officials say this is illegal and that they are only asking for a license to use the AI for lawful activities.
However, a senior Pentagon official claimed that the main concern was not mass surveillance, but rather ensuring human involvement in final targeting decisions. They argued that the government should have control over how its technology is used, just like with Boeing planes.
Due to lack of trust between the two parties, officials are considering using the Defense Production Act to exert control over Anthropic and may even designate the company as a “supply chain risk.” This could push Anthropic out of government contracts. The company had been authorized to work on classified networks and is committed to supporting national security, according to their statement.
The situation is developing quickly, with Hegseth’s deadline for access to Claude looming by Friday.
Source: https://www.cbsnews.com/news/hegseth-anthropic-full-access-claude-ai-model