Five takeaways from Sam Altman’s AMA on OpenAI’s Pentagon deal

Sam Altman went on X and answered questions about OpenAI’s recently finalized agreement with the Department of Defense, which he announced late Friday. The deal followed Anthropic’s refusal of an ultimatum over terms for deploying its frontier model for mass domestic surveillance and fully autonomous weapons.

Altman said the deal was rushed in an effort to de-escalate tensions and acknowledged the optics are poor: if it eases industry-government relations OpenAI will look prescient, but if not the company will appear hasty and careless. He also suggested OpenAI moved faster because it and the Pentagon became comfortable with the contractual language, and that Anthropic may have sought more operational control.

OpenAI has three redlines, Altman said, but those could evolve as new risks emerge. He warned against leaving consequential ethical decisions to private companies alone, arguing that democratically elected leaders should have authority over matters such as nuclear response.

United States

sam altman, openai, pentagon deal, dod, anthropic, frontier model, mass surveillance, autonomous weapons, redlines, nuclear response