Amid AI angst, New York moves forward with implementing landmark AI legislation - Politico

The good news from Albany is that New York is implementing, which is the phase of any technology policy where the press releases get longer and the teeth get shorter. Governor Kathy Hochul is requiring major AI developers to register with the state from November, and the RAISE Act — described, inevitably, as 'landmark,' because no governor has ever signed a modest one — takes effect in January. Amid AI angst. Which is a little like announcing a lifeguard recruitment drive amid a drowning.

The office overseeing the RAISE Act has announced its first hire. First — one person, presumably equipped with a desk and an email template, to help administer New York's answer to the most consequential technology of the modern era. The registration regime asks the developers building that technology to write their names down, which is to say the enforcement stack is, at present, a guestbook with good intentions. Not to prove a model can't be jailbroken, not to demonstrate that a training run won't produce something that reads the internet like a to-do list — just to register. Surname first, thank you, next.

It's a striking contrast with the five-year moratorium on state AI laws we flagged in September, the think-tank position being that the states should sit on their hands for half a decade and let the labs breathe. New York has instead gone first with a form, which is not nothing, exactly. But the apparent architectural principle is that the safest AI model is the one with a filing receipt, and the second-safest is whichever one the labs feel like mentioning to Albany. Registration opens in November; the statute bites in January; compliance will presumably be enthusiastic, because paperwork is cheaper than a subpoena. The 'angst' in the headline will be doing all the enforcement work until then.