New York used AI to scan every state rule. The $25 dog hunting fee didn't stand a chance

New York Governor Kathy Hochul just signed a moratorium on new AI data centers in the state, but she is not against using the technology herself. During an interview with Bloomberg's Odd Lots podcast, Hochul said her team used "AI to analyze every single rule, regulation, [and] policy" to check for outdated legislation. The moratorium affects exactly the kind of infrastructure that makes AI possible, a tension worth watching, given demand for AI compute is far from done growing.

Some of the antiquated laws Hochul mentioned include a $25 fee required to take a dog hunting, or a stipulation that pregnant people need a permit to work after midnight. She added that it "probably would have taken five years at the staff level" to review all the laws. With AI, her team "did it in a couple of months." That will let Hochul and state agencies "get rid of" these outdated regulations. The productivity gain mirrors a pattern in the private sector, where the hardest problem in AI coding agents isn't code, it's counting the hours saved for humans.

Earlier this week, New York became the first state to pause new hyperscale data centers for up to a year. In the meantime, lawmakers plan to regulate to protect residents from rising utility costs and threats to natural resources posed by massive data centers. This is the double-edged nature of AI in government: the technology that enables efficiency also demands the energy and water that regulators now must limit.

"I want a government that's not on your back but on your side, and using AI has been powerful to do that," Hochul told Bloomberg. "I think every level of government should use this. I'm going to make dramatic changes using the power of AI." Whether that power gets directed at cutting red tape or managing the infrastructure that makes AI possible will define the next few years for states like New York. The use case also raises questions about how AI audits of government regulations compare to similar tools in other domains, such as OCR models that still beat newer rivals on specialized language tasks.