New York's government is enlisting artificial intelligence to do something few administrations have tried at scale: read through the entire body of state regulations and flag rules that may be outdated, duplicative or unnecessarily burdensome.
Governor Kathy Hochul has described the effort as a sweeping "regulatory reset," using AI tools to help analyse the state's rulebook and identify candidates for simplification or repeal. The stated goal is to reduce the compliance costs that regulations impose on businesses and residents, and to speed up government processes the administration says have been slowed by accumulated red tape.
How it is meant to work
The idea is not to hand decisions to a machine. In the administration's telling, the AI acts as a filter: it scans the vast, dense text of state regulations, much of it built up over decades, and surfaces rules that appear obsolete or redundant. Human officials and agency experts then review those suggestions before anything is actually changed, the state says, so that the final judgement rests with people rather than software.
Governments everywhere struggle with regulatory clutter, layering new rules on top of old ones without removing what has become irrelevant. Proponents argue that AI is well suited to the tedious work of combing through thousands of pages to find contradictions and dead letters that human reviewers might never have time to catch.
Promise and pitfalls
The appeal is obvious. If it works, the approach could make government leaner and less costly to deal with, and free officials from paperwork that serves no one. Supporters frame it as exactly the kind of unglamorous, practical task that AI can genuinely help with.
But the plan also draws caution. Critics of deregulation more broadly worry about what gets swept away in the name of efficiency: a rule that looks outdated to an algorithm may in fact protect workers, consumers or the environment. Others raise questions about transparency and accountability, how the AI reaches its conclusions, how errors are caught, and who is answerable if a needed safeguard is wrongly discarded. The reassurance that humans remain in the loop is central to the administration's case, and how rigorously that review is carried out will determine whether the exercise is prudent housekeeping or a shortcut with real risks.
Part of a bigger question
New York's move lands amid a broader debate about how governments themselves should use AI, even as they weigh how to regulate it. Deploying the technology inside the machinery of the state raises the same issues officials are grappling with elsewhere: accuracy, oversight, and the temptation to trust automated judgements more than they deserve.
For now, the initiative is a notable experiment in applying AI to the business of governing rather than merely policing it. Whether it becomes a model for other states or a cautionary tale will depend less on the cleverness of the software than on the care with which its recommendations are checked, and on whether the rules it helps remove turn out to be genuinely useless, or quietly important.
