AI Giants Are Fighting Over Whether Anyone Should Police Them

For a few days this week, it looked like the biggest names in AI had finally agreed on something: that artificial intelligence might be moving too fast and needs some kind of referee. Anthropic CEO Dario Amodei floated a three-step plan — let outside experts inspect AI labs, get rival companies to coordinate, and eventually sign international agreements. OpenAI's Sam Altman and Google DeepMind's Demis Hassabis seemed to nod along. Parts of the industry even discussed creating an independent watchdog, similar to how the financial world has FINRA, a private body that polices stockbrokers.

Then the pushback started. Meta CEO Mark Zuckerberg said each lab should decide for itself how fast and how safely to build. The Wall Street Journal reported that Zuckerberg, Elon Musk and Nvidia CEO Jensen Huang helped scuttle the independent regulator idea. President Trump weighed in too, calling recent AI safety fears a "hoax" and joking that the only guardrail AI needs is a "strong and smart" president. So the industry that spent months warning about dangerous AI is now publicly split over who, if anyone, should be allowed to slow it down.

Why should you care? Because this fight decides the rules you'll live under. If AI runs your health insurance claim, screens your job application, or writes your kid's homework, someone has to decide what it's allowed to do — and how you'd know if it went wrong. Right now, that someone is mostly the companies themselves, and they don't agree on much.

There's a catch in both camps' arguments. The pro-regulation labs say they want rules, but they're also the ones who'd help shape them — critics call that "regulatory capture," where the regulated write their own rulebook. The anti-regulation camp says companies have every incentive to be safe, but incentives don't help you if a model leaks your data or a biased hiring tool rejects you. Either way, expect this argument to keep changing week to week.