Anthropic CEO Dario Amodei criticized OpenAI’s Pentagon partnership in a leaked internal memo. Amodei labeled OpenAI’s messaging as lies and described the deal as 80% safety theater. The memo accuses OpenAI of undermining safety principles and misrepresenting contractual safeguards to the public.
These comments followed the breakdown of Anthropic’s own negotiations with the Pentagon. Anthropic refused to remove safeguards preventing AI use for mass domestic surveillance. The company also rejected terms allowing its technology to power fully autonomous weapons.
The U.S. government designated Anthropic a supply chain risk following the negotiation failure. OpenAI announced its Pentagon agreement hours after the government issued that designation. Amodei has since apologized for the tone used in the leaked memo.