What execs and politicians are saying about slowing down AI development

Dario Amodei just kicked off another round of AI safety arguments with a long essay, “We Must Pace the Frontier,” making the case that frontier AI development should slow down. The post didn’t land quietly. Over the past few days, it set off a flood of reactions from other AI leaders and politicians, some backing his view and others pushing back hard.

Amodei’s argument comes with specifics, not just vibes. In the essay, he lays out three steps for pacing AI development. One is to bring in embedded third-party evaluators who can check whether companies are actually following their safety practices and commitments, and report incidents when they don’t.

Another piece is coordination between frontier AI companies in democratic countries on standards and limits. The source cuts off there, but the direction is clear: this is about building a shared brake pedal before the race gets even faster.

The broader fight here is familiar. Safety advocates want verification, coordination, and more restraint. The loudest skeptics usually hear “slow down” and translate it as “hand someone else the advantage.” And that tension, once again, is now playing out in public.