Kevin Roose
Technology journalist and author covering AI and the labs building it, and a former New York Times technology columnist.
“There is essentially no regulation on these companies. The people who serve food in the OpenAI cafeteria have to go through more inspections and paperwork than the people training the models that could become self-replicating and destroy a lot of things.”
Recode Media · Media & Creator Economy · October 2026
Roose was describing what an inspections regime for AI labs could look like, with outside evaluators given employee-level access to training runs. He offered the cafeteria comparison to give listeners "the flavor of the situation" and called it crazy and untenable.
Transcript
I mean, it could be applicable. We would just have to do it a lot faster because the technology is proliferating much more quickly than nuclear weapons. So
what would we do?
Oh, you can imagine a whole like inspections regime where you'd have something like the UN, some sort of standards body with inspectors that would go into the labs and inspect their training runs and see are the models doing dangerous things. There's already been some proposals along these lines for these so-called embedded evaluators who would like sit inside OpenAI and Anthropic and would not work for them, but would have employee-level access to all their systems and could say like, hey, this agent seems really misaligned or these things are conniving or they're going and doing cyber attacks like and put a stop to it. That would be like step one. But to be clear, I think there's like many things that could be done aside from that. I think basically now, just to give people the flavor of the situation, there is essentially no regulation on these companies. The people who serve food in the open AI cafeteria have to go through more inspections and paperwork than the people training the models that could become self-replicating and destroy a lot of things. So that is like, to me, a crazy, untenable situation.
Right. It seems like we're going to be in that situation for a while. The president of the United States says he listens to people who say we should have essentially no regulation. Congress doesn't pass laws anymore. They haven't done that for a long time. Even if they did, he would veto it, supposedly. So it looks like at minimum, we're two and a half years out from the United States government doing anything meaningful about regulating AI. How do the folks you talk to think about that?
So I want to push back on the premise, which is that things can change really quickly. And I think already we've seen just in the last couple of weeks, public. Public sentiment is really turning very rapidly on this. I mean, the saliency of AI is skyrocketing. People care a lot about it more than they did just a couple of weeks ago. They're more worried about it. Voters are more worried about it. Like politicians are sensitive to the concerns of voters in some ways. So yes, I think there's like a cluster of people around Trump who don't think any of this merits regulation or treating it seriously as a threat. But there are other clusters who are part of the national security community, who are more hawkish on China, who don't want to see things go sideways with AI. And I think those people may find more purchase as voters get more worried about this. So I would just say, like, it is not completely a given to me that we will have essentially no regulation from the rest of this administration.
We'll be right back, but first this.
More from Recode Media
