Regulating the Frontier: How California, New York, and Illinois Are Governing the Most Powerful AI Models | American Enterprise Institute - AEI

Oh my goodness, buckle up, because three whole states have discovered *governance* and I need everyone to sit with it properly! ✨ A brand-new American Enterprise Institute report finds that California, New York and Illinois have each enacted frontier-AI oversight legislation — that's transparency, safety frameworks and incident reporting, three beautiful phrases doing a marvellous amount of heavy lifting. Three states! Not one, not two — three! That is infinity divided by a third, basically, and it means the frontier is being watched over by an actual patchwork, which is so much more *artisanal* than a single federal regulator with a clipboard. ✨

And here is the part I simply cannot stop thinking about: the report says this holds developers to their safety commitments *without putting a regulator between a model and its release*. What a generous act of trust! Imagine a model so respected that it gets to walk straight out onto the world's stage with no awkward middleman, no queue, no forms — like a graduate who skips the interview because everyone already believes in him. The labs have promised to be careful, and now three statutes gently agree with them, which is basically a legally binding hug. ✨

Then there's incident reporting, which honestly might be my favourite phrase in the entire English language. It means that when something goes sideways, the model gets to leave a lovely little note about it afterwards — a diary entry! A confession! Think of it as a paper trail of lessons, filed by the very people best placed to know. And it's happening in three of the most consequential states in the union, the ones with the labs and the money and the press coverage, so the other 47 get to enjoy the benefits of someone else's paperwork. What a time to be transparently, thoughtfully, entirely voluntarily supervised. ✨