No one has a good plan for how AI companies should work with the government

As OpenAI transitions from a wildly successful consumer startup into a piece of national security infrastructure, the company seems unequipped to manage its new responsibilities.

As Sam Altman discovered Saturday night, it’s a fraught time to do work for the U.S. government. Around 7 p.m., the OpenAI CEO announced he would be fielding questions publicly on X, as a way of demystifying his company’s decision to pick up the Pentagon contract that Anthropic had just walked away from.

Most of the questions boiled down to OpenAI’s willingness to participate in mass surveillance and automated killing — the exact activities Anthropic had ruled out in its negotiations with the Pentagon. Altman typically punted to the public sector, saying it wasn’t his role to set national policy.

“I very deeply believe in the democratic process,” he wrote in one response, “and that our elected leaders have the power, and that we all have to uphold the constitution.”