Is Big Tech’s AI slowdown a safety pact or a cartel?

Over the weekend, Sam Altman, Dario Amodei, Demis Hassabis, and Elon Musk loosely agreed that AI development should slow down. The message they attached to it was neat and public: they said they wanted to “pace the frontier.”

They also signed on, at least in part, to a proposal built around third-party auditors, rules for domestic labs, and some kind of global slowdown agreement. That sounds careful on paper. It also sounds like the sort of thing that can be used to decide who gets to move and who gets left waiting at the gate.

Not everyone heard “safety” in the announcement. Skeptics immediately argued that the real goal was to make life harder for would-be competitors, especially the open-source crowd, while sidestepping stronger legal safeguards that would actually bite. Some went further and called it a cartel.

That accusation sticks because the people involved are not randos. They are the executives and founders who sit closest to the center of AI power. When they start talking about slowing the whole field down, it is fair to ask whether the brakes are for everyone or just for the people trying to catch up.