The situation
New York Governor Kathy Hochul says her team used AI to review every law on the state’s books. She told Bloomberg’s Odd Lots podcast that the team deployed “AI to analyze every single rule, regulation, [and] policy” to hunt for outdated legislation, according to The Verge AI. The job took a couple of months. Done by staff, Hochul estimated it “probably would have taken five years.”
Same week, she signed a moratorium pausing new hyperscale data centers in the state. Hold that thought.
Intel summary
- Scope: Every rule, regulation, and policy in New York State. Not a sample. The whole corpus.
- Targets found: A $25 fee required to take a dog hunting. A stipulation that pregnant people need a permit to work after midnight. Fossils, still technically enforceable.
- Time compression: Five years of staff work down to roughly two months.
- Next move: Hochul says she and state agencies will now “get rid of” what the review flagged.
- The quote that matters: “I want a government that’s not on your back but on your side, and using AI has been powerful to do that. I think every level of government should use this. I’m going to make dramatic changes using the power of AI.”
Why this one lands
What stands out here is how unglamorous it is. No chatbot mascot. No copilot in a press release. This is document review at a scale humans physically can’t cover: read a massive corpus, classify what’s stale, surface it for a decision.
That’s the most defensible AI use case in government, and it’s been sitting there the whole time. States run on decades of accumulated text nobody has read end to end since it was written. Rules pile up. Nobody gets promoted for deleting them. The backlog isn’t a policy problem, it’s a reading problem, and reading at volume is exactly what these models do well.
Before this, deregulation efforts meant hiring a commission, giving it two years, and getting a partial report. Hochul just changed the unit economics of that whole exercise.
The contradiction nobody’s ignoring
New York became the first state to pause new hyperscale data centers, for up to a year. State lawmakers plan to use that window to write rules protecting residents from rising utility costs and the strain on natural resources, per The Verge AI.
So: pause the buildout, use the output. Both positions can hold at once. You can find AI useful and still think a 500-megawatt campus shouldn’t land next to a suburb without a plan. But the tension is real, and Hochul will get asked about it every time she repeats the “dramatic changes” line.
What to expect next
- Copycats: Governors watch each other. This is cheap, easy to explain, and hard to attack from either party. Expect two or three states to announce something similar within the year.
- A vendor land grab: Regulatory corpus review is a real product category now, and the reference customer is a US state.
- The bottleneck moves: Finding a dead rule takes a model. Repealing it takes a legislature. Hochul’s team has a list. Lists have died in committee before.
- The verification question: Nobody’s said how the flagged rules were checked, what the false positive rate was, or how much human review sat on top. That’s the detail that decides whether this scales or becomes a cautionary tale.
The read
This is significant because it moves AI in government from pilot theater to a shipped result with a number attached. Five years to two months is the kind of claim other officials can put in a budget request.
It’s also a reminder of where the real value sits right now. Not in the flashy demo. In the pile of text nobody wants to read.
The test comes next: whether those outdated rules actually get repealed, or whether the list just becomes another document sitting on a shelf. Full details are at the original source.