New York's state government just completed what it describes as the largest regulatory review in its history. Normally, this kind of top-to-bottom review of every law, rule, and policy across dozens of agencies would take years of staff time. Instead, it was done in a couple of months, using an AI system built by Stanford University's RegLab, a research group that has spent years building data tools specifically for government use.
The AI flagged thousands of rules as candidates for reform. The state then had officials and outside experts review those flags before acting on them. That combination, machine speed plus human judgment, is the right way to use this kind of tool. The AI does not make the decisions; it just does the reading at a scale no team of people could match.
Some of the results are almost comic. A $25 fee to take a dog hunting. A requirement that pregnant people obtain a permit to work after midnight. These things exist because rules get written and then simply forgotten. Nobody goes back to check. The AI went back and checked everything.
The practical value is bigger than the odd examples suggest. The first package of changes covers 50 regulatory actions across 22 agencies. The state estimates these will save New Yorkers tens of millions of dollars in fees and compliance costs, and more than a million hours of paperwork annually. Over 1.5 million residents are expected to benefit from just this first wave. The changes include simplified license renewals for more than 800,000 licensed professionals, faster affordable housing reviews, and reduced paperwork for healthcare providers.
The public also contributed. Around 4,000 New Yorkers submitted ideas through a public comment process that ran earlier this year, suggesting rules they wanted gone. So the review was not purely algorithmic. It combined AI scanning, expert review, and direct public input.
There is a broader context worth noting. Hochul signed this initiative at the same time New York became the first state in the country to impose a one-year pause on building new large-scale data centers. The pause applies to facilities that use 50 megawatts or more of power, which is the scale required to run serious AI training operations. The stated reason is rising electricity bills and strain on natural resources. The move drew sharp criticism from President Trump, who called it a terrible decision.
So Hochul is doing two opposite-looking things at once: cutting the state's own red tape with AI, while also restricting the physical infrastructure that makes AI possible. That is not necessarily a contradiction. It reflects a political calculation that the benefits of AI software tools are popular and visible, while the costs of AI infrastructure, higher electricity bills, water use, grid stress, fall on ordinary residents and voters.
For anyone running an organization that operates across multiple states or interacts with government licensing, procurement, or permitting, this matters in a concrete way. If this approach spreads, and it likely will, the regulatory environment your business operates in could shift faster than it has in decades. Rules that took years to change could be updated in months. That is good for businesses trapped under outdated requirements. It also means the window for adaptation gets shorter.
Stanford's RegLab has worked with the EPA, the IRS, the Department of Labor, and local governments before New York. The tools exist and have been tested. The New York exercise is the largest deployment of this kind at the state level, and other governors will be watching the results closely.