2025 AI Policy: 85% of Cities Lag Behind

A 2025 report from the Center for AI Policy reveals a startling fact: less than 15% of American cities have initiated formal discussions about local AI policy. This is happening even as AI is being rapidly deployed in urban infrastructure and public services. That figure isn’t just a number. It points to a critical blind spot, especially for a hub like New York City, which is dealing with its own unique set of problems and opportunities with this tech. How NYC handles local AI policy could very well become the template for how other cities manage the tricky dance with incoming federal AI regulation.

2025 Data: Under 15% of US Cities Address AI Policy

That statistic, that under 15% of US cities have begun formal discussions on local AI policy, is a flashing red light for anyone in urban planning. To me, this shows a massive governance vacuum. AI is already here, embedded in everything from predictive policing algorithms to smart city infrastructure. Right now, in 2025, cities like Los Angeles and Chicago are still mostly just reacting to AI incidents instead of building proactive frameworks. That reactive stance just creates a messy patchwork of regulations, which breeds inconsistency and kills public trust. New York City, with its density and tangled tech dependencies, simply can’t afford to be that slow. Without clear local guidelines, you get unregulated AI deployments that can easily make existing inequalities worse, violate privacy, and fail to deliver on their promises. This includes everything from automated decision-making in housing applications to AI-powered surveillance in public spaces like Times Square. Without local policy, these systems are operating in a regulatory gray zone, leaving residents exposed and city agencies with no clear direction.

NYC’s 2024 Algorithmic Transparency Bill: A Local Precedent

Going against the national trend of inaction, New York City passed an algorithmic transparency bill in 2024 that requires city agencies to disclose information about the AI systems they use for public services. This legislation, enacted by the City Council and overseen by the Mayor’s Office of Technology and Innovation, shows a city trying to get ahead of the problem. My take is that this bill, while far from perfect, is a necessary first step toward building some public trust and accountability. It forces agencies to actually think through the ethical side of their AI tools and gives the public a mechanism for scrutiny. For instance, when the Department of Social Services uses an AI tool to assess benefits eligibility, the new law mandates clear explanations of how the AI made its call. This is miles ahead of federal efforts, which, as of 2026, are still focused on broad ethical principles and not concrete implementation requirements. The federal government, through things like the National Institute of Standards and Technology’s (NIST) AI Risk Management Framework, provides useful guidance, but it has no real enforcement power over city operations. NYC’s bill, which also includes provisions for public comment periods before new AI systems are rolled out, shows how local policy can be more agile and responsive to community concerns than a top-down federal directive.

30% of NYC Public Services Now Incorporate AI

Here’s a staggering number: a 2025 audit by the city’s Comptroller’s Office found that 30% of New York City’s public services now incorporate some form of AI. This isn’t theoretical. It covers a huge range of applications, from optimizing sanitation routes in Queens to improving emergency response times in the Bronx. This widespread adoption shows just how urgently a complete local AI policy is needed. This statistic reveals both the huge potential and the serious risks of AI in cities. On one side, AI can make services much more efficient. You could have systems that predict infrastructure failures on the Brooklyn Bridge before they happen or that optimize traffic on the FDR Drive during rush hour. But on the other, the sheer pervasiveness of AI in essential services requires strong oversight. We’ve already seen cases in other cities where biased algorithms led to discriminatory bail recommendations or parole decisions. Without clear local guidelines on data privacy and algorithmic fairness, NYC risks embedding systemic biases directly into its public infrastructure. Given the volume of AI integration, even a small flaw in design can have a massive, negative impact on millions of people. The city’s Office of Technology and Innovation has been developing internal guidelines, but those often don’t have the legal teeth of formal legislation.

Federal AI Framework: Broad Strokes, Local Gaps

The federal government’s main strategy for AI governance is the 2023 Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence. It provides a foundation, but the framework mostly offers broad guidelines that leave significant gaps for local implementation. The federal strategy is, by necessity, a high-level one that aims to encourage innovation while managing big national security and societal risks. It addresses grand challenges like safeguarding the country and ensuring responsible development. What it doesn’t do, and probably can’t do, is get into the granular details of AI deployment in a city as complex as New York. Take the federal government’s emphasis on “equity and civil rights.” That’s a great goal, but it needs to be translated into specific, enforceable policies at the city level. How exactly does a federal mandate on equity apply to an AI system the Department of Buildings uses for permit approvals in different neighborhoods with their own unique socio-economic profiles? The feds aren’t going to dictate NYC’s data collection practices or set the parameters for algorithmic impact assessments within local agencies. That’s where local policy must come in. The federal framework is like the interstate highway system. It connects major points, but cities still need to build their own intricate network of local roads and traffic laws. This difference creates a chance for cities like NYC to lead on governance, but it also risks creating a disjointed mess if local and federal work isn’t coordinated.

The Conventional Wisdom: Federal Preemption is Inevitable

A lot of policy experts assume that federal preemption in AI regulation is inevitable. Their argument is that since AI is a borderless technology, national standards are the only way to get consistency and prevent a “race to the bottom” in oversight. I strongly disagree, at least in the near term. While federal involvement is obviously needed for things like national security and broad ethical principles, the idea that a single federal framework can effectively govern every AI application in thousands of different municipalities is just not practical. Think about it: the unique challenges of AI in the New York City subway system (like predictive maintenance for old infrastructure and crowd management) are worlds apart from those of a rural farming community using AI for crop optimization. A federal agency, no matter how well-intentioned, can’t possibly write regulations that are equally effective for both. Local governments, with their direct knowledge of community needs and existing infrastructure, are in a unique position to craft nuanced AI policies. For example, data privacy concerns around facial recognition in public spaces are far more acute in a dense urban center like Manhattan than in a sprawling suburb. Local officials understand these differences. Expecting a single federal mandate to cover all those local details is a fantasy. Instead, what we’ll likely see is a layered approach, with federal guidelines providing a broad ethical ceiling and local policies providing the specific, enforceable rules that govern daily AI interactions. This is about federal *and* local, working together. The strength of NYC’s approach, which is part of the ongoing conversation around the city’s role in technology piloting, is its ability to tailor solutions to its specific urban fabric. This all points to a fundamental truth: technology must serve the people it impacts. Local policy, grounded in community needs and implemented with rigorous oversight, is a necessity for ensuring AI’s benefits are realized ethically and equitably.