The plan to pace the AI frontier runs every safeguard through the labs themselves

A team that has standardised on a frontier model reads a post like this looking for a date and a party who can be held to it. Neither is in what the Equity hosts were able to describe. Sean O'Kane said the plan is short on specifics, both on why these executives believe the risks are there and on what they mean by slowing down ^[5].

Kirsten Korosec named three pieces: independent third-party evaluators working inside companies like Anthropic or OpenAI to monitor safety practices and incidents ^[8], coordination on safety standards and limits among major AI companies in democratic countries, and an international coordination piece ^[9]. Three for three, the party doing the checking is the labs and the evaluators the labs agree to seat ^[16]. Korosec said there is a lot of vagueness around the evaluator proposal ^[8].

O'Kane was blunt about customer pressure. "There does not seem like there is a ton of consumer choice driving this market, in the sense of, okay, if one of these companies does something really bad, then they will see an impact in how many cancellations there are or whatever," he said ^[7].

On the regulatory side he was plainer. "We have a federal government at the moment that certainly does not seem eager to enforce regulations broadly speaking, let alone specifically speaking here," O'Kane said ^[6].

Then there is the speed of the agreement. Anthony Ha said he saw other people posting some version of the plan in the days before Amodei published it, and that the underlying suggestions came out of the AI safety community, where they had been percolating for a while ^[12]. Korosec set both readings of that side by side. "First of all, consensus can be a good thing. Also, it can be the beginnings of forming a cartel," she said ^[10]. She also narrowed the scope to frontier AI labs, which she called "specific and important" ^[11].

The loudest objection is to regulation itself. Nvidia's Jensen Huang has echoed President Trump's claim that the AI backlash is a hoax and that regulation is unnecessary ^[2], and Korosec pointed to the call he took from Trump on stage at the All-In Summit ^[15].

Two questions decide whether any of this should touch a roadmap: who verifies the commitment, and what the remedy is when it lapses. For all three pieces described on the episode, both answers come back to the labs and the evaluators they choose ^[16]. O'Kane had said a week earlier that he did not think a slowdown of any kind was coming, in large part because he does not think these companies are structured in a way where that works ^[13].