Zuckerberg just broke ranks with three AI CEOs.

On Saturday, Anthropic's Dario Amodei published an essay arguing the industry must slow the rate at which it improves model capabilities.

Sam Altman agreed within hours. Elon Musk posted "Dario is right."

On Tuesday, Zuckerberg said no.

Here's what most of the coverage is getting wrong.

He isn't arguing for speed. Meta held its Muse models back for months over safety, and noted they didn't ask anyone else to do that first. He also backs independent evaluators, which is Amodei's own first proposal.

The disagreement is narrower than that. And bigger.

Amodei wants labs coordinating with each other, shared checkpoints, governments eventually involved.

Zuckerberg wants every lab braking on its own evidence, its own customers, its own legal exposure.

The fight isn't whether to slow down. It's who holds the brake.

That decides what building on these models feels like downstream. Shared standards and slower capability gains, or faster models where you carry the risk yourself.

I've been building AI products since 2014, and two weeks ago none of this was being argued in public.

Now the four biggest labs are doing it out loud, and three already agree on letting outsiders inside.

Who should decide when a lab is moving too fast?