Anthropic Doesn't Want Its AI Killing People. The Pentagon Isn't Happy.
Posts from this topic will be added to your daily email digest and your homepage feed.
TL;DR
- Anthropic is in a public dispute with the Pentagon regarding its AI's acceptable use policy.
- The Pentagon is pushing for "any lawful use" of AI services, which Anthropic believes could lead to mass surveillance and lethal autonomous weapons.
- The Pentagon has threatened to designate Anthropic as a "supply chain risk," potentially jeopardizing its $200 million contract and relationships with other defense contractors.
- Anthropic's CEO, Dario Amodei, is set to meet with Pentagon CTO Emil Michael to negotiate the terms.
- Anthropic's refusal stems from its policy against autonomous kinetic operations and mass domestic surveillance, citing concerns over civil liberties and the current state of technology.
- Existing government directives support Anthropic's position on human judgment over the use of force and restrictions on collecting information on U.S. persons.
- The Pentagon, under Secretary Pete Hegseth, is prioritizing speed and an "AI-first" warfighting force, potentially overlooking safety and alignment risks.
- Other AI companies like OpenAI and xAI have reportedly agreed to the "any lawful use" terms, but their models may not meet the necessary security classifications to replace Anthropic's Claude.
- Anthropic's Claude is currently the only frontier AI model operating on fully classified Pentagon networks.