> ## Content Index
> Fetch the complete content index at: https://www.vitalcitynyc.org/llms.txt
> Use this file to discover other available public pages before exploring further.

# City Limits
- URL: https://www.vitalcitynyc.org/federal-ai-regulation/
- Published: 2026-11-18T17:00:00.000Z
- Updated: 2026-11-18T17:00:00.000Z
- Description: Why Washington, not New York City (or state), should be the frontline regulator of AI
- Author: Kevin Frazier
- Tags: #ai-and-the-city, Artifical Intelligence, Federal Government, Technology, Retraining Our Systems

Since the nation’s founding, municipalities have been [carefully crafted to identify local issues](https://academic.oup.com/ajlh/article-abstract/20/2/171/1802834?ref=vitalcitynyc.org) and develop solutions. A Control-F search of the U.S. Constitution for “city,” “municipality” or even “local” will return zero hits. Omission should not be conflated with insignificance. Cities have become an essential vehicle for democratic discourse and collective action — filling gaps left by states and the federal government and facilitating international coordination on a massive scale. Today’s mayors may go from assessing pothole repair plans to brokering climate accords with counterparts from around the world. It’s unsurprising that many now have leading AI companies, such as OpenAI and Anthropic, on their regulatory agendas. Yet, the temptation for city officials to venture into AI governance ought to be checked based on legal constraints and institutional limitations.

The founding generation viewed local governance as a hyperfunctional matter. State legislatures granted corporate charters to cities to hold, distribute and regulate property. This incorporation process specified the powers afforded to city officials, how such officials would be hired, fired and replaced, and which properties and processes they could regulate. 

As city governments evolved — shedding corporate formalities and embracing more democratic processes — they have drifted from narrowly focusing on enumerated issues to tackling more abstract issues such as the [general welfare and public safety](https://uncpress.org/9780807846117/the-peoples-welfare/?ref=vitalcitynyc.org) of their residents. This expanded authority was the product of law and culture. Rather than being granted special charters with fixed powers, cities soon received general charters from state legislatures — freeing them to address a broader range of issues. 

Contemporary American cities have amassed extensive policy portfolios. In addition to managing the issues set forth in the charters of yesteryear — maintaining local infrastructure, putting out fires, cleaning up trash, etc. — major cities have taken on a novel set of tasks, including assistance with international diplomacy. [More than 350 cities](https://www.climatemayors.org/?ref=vitalcitynyc.org), for example, have joined Climate Mayors, a network tasked with aligning urban climate policy. 

While some have applauded this expanded role as necessary given political friction at other levels of government, others have derided a loss of focus on what is squarely in the ambit of cities. Los Angeles [Mayor Karen Bass](https://abc7.com/post/mayor-karen-bass-admits-ghana-trip-before-wildfires-was-mistake/15929179/?ref=vitalcitynyc.org) being absent from her city during horrendous wildfires due to a trip to Ghana on behalf of the Biden administration is perhaps the most well-known case study in the pitfalls of city officials emulating national politicians. Tasked with doing everything, they may not properly identify the issues best suited for their attention based on their legal authority, resources and popular mandate.

What role, then, should cities play in AI governance?

Forecasts of AI causing widespread, immediate and long-lasting damage have unsurprisingly caused a [demand for regulation](https://theaipi.org/poll-ai-safety-majority/?ref=vitalcitynyc.org) by the general public. Many of those scenarios involve risks that will manifest at the local level: high levels of joblessness, increased surveillance and a broken information ecosystem that erodes trust. 

Tellingly, however, when [pollsters ask](https://theaipi.org/poll-ai-safety-majority/?ref=vitalcitynyc.org) [which entity](https://hai.stanford.edu/ai-index/2026-ai-index-report/public-opinion?ref=vitalcitynyc.org) the public wants to regulate AI, they omit local government as an option. The implied conclusion is that cities are not in a position to solve loss-of-control scenarios in which AI tools escape their confines and wreak havoc on unsuspecting third parties or the geopolitical tensions around who can develop frontier AI systems, when and to what ends. 

That’s likely the correct understanding and has been announced by mayors from [Tim Kelly in Chattanooga](https://www.congress.gov/119/meeting/house/118288/documents/HHRG-119-IF17-20250521-SD003.pdf?ref=vitalcitynyc.org) to [Zohran Mamdani in New York](https://www.politico.com/news/2026/10/05/in-nyc-mamdani-cedes-ai-regulation-lane-01108016?ref=vitalcitynyc.org). That said, it’s not a uniformly held view among city officials, including Julie Menin, speaker of the Council of the City of New York, [who called for OpenAI’s Sam Altman and Anthropic’s Dario Amodei to join in the Council’s investigation of potential risks of frontier AI](https://www.6sqft.com/city-council-rare-special-hearing-on-ai-risks/?ref=vitalcitynyc.org).

If AI stakeholders are already worried about 50 different state policies complicating their ability to comply with the law, it’s hard to imagine even the largest AI labs being able to navigate hundreds of different AI specifications implemented by cities. Beyond that practical concern, there’s also a constitutional question of whether cities have the legal authority and public legitimacy to exercise control over systems used by billions of people. 

> A resident cannot audit the Commerce Department, but she can walk into city hall, attend a council meeting and judge for herself whether the permit office got faster or the pothole got filled.

Cities were not intended to be the first line of defense in matters of national and international significance. When mayors step into the shoes of governors or even senators, they meddle in a system that’s calibrated to ensure the public can hold actors accountable for specific tasks. 

So mayors and city councilors, even those in large cities, should not set their sights on regulating the frontier of AI. Matters such as when and how AI is developed rest with the federal government as the body assigned to matters of national significance. Likewise, even questions of how AI should be deployed in public and commercial settings are in many cases outside the core responsibilities of cities. 

Of course, the failure of one level of government to address issues in its domain invites participation from other government actors. When the stakes are high (and they are in the AI space) and public attention is at a fever pitch (and it is), then the urge for any government actor to step in becomes nearly irresistible. 

Yet, it’s key that local governments use their sparse resources to respond to the issues squarely in their domain because it’s unlikely and arguably impossible for states or the federal government to step into those hyperlocal, complex policy questions. These tasks may not garner headlines, but they are necessary to building more resilient communities amid disruptive developments in AI. 

A simple framework can assist with distinguishing between the AI policy tasks aligned with city capabilities and duties and those that are best left to another level of government. 

First, is it a policy issue that is directly and uniquely impacting residents of a specific community? For the sake of illustration, application of this principle steers local officials away from trying to develop high-level policy about the economic implications of AI, such as a potential increase in income inequality, and instead sets their gaze on the local manifestations of economic disruption, such as how to meet the hiring needs of local employers looking for employees with new sets of skills. 

No two communities will experience the same economic consequences from the AI buildout. The significant layoffs in the gaming industry — attributed by many to AI — were [acutely felt](https://spectrumlocalnews.com/ca/california/entertainment/2026/07/31/artificial-intelligence-video-games?ref=vitalcitynyc.org) in Southern California. Local officials there are in a position to rally other employers, workforce training entities and education institutions to help support individuals who may be affected by AI-related layoffs (or layoffs resulting from any other range of factors). Now’s the time to develop workforce disruption plans so that changes already manifesting in specific sectors and among entry-level workers do not result in any more negative consequences than necessary. Second, does the policy implementation require extensive participation from local actors, including but not limited to public entities? 

A federal plan for increasing AI literacy may include considerable financial resources but will run aground if it does not include local buy-in. [More than any other level](https://news.gallup.com/poll/512651/americans-trust-local-government-congress-least.aspx?ref=vitalcitynyc.org) of government, the public trusts local officials, albeit with a major qualification: 67% of Americans have a great deal or fair amount of confidence in “local governments in the area where you live *when it comes to handling local problems*.” Local officials can leverage that confidence to empower trusted messengers and institutions to address some of the trickier issues on AI governance.

Imagine, for example, city officials turning public libraries and community centers into AI demonstration centers — spaces where people can learn about AI for free and without corporate influence. 

Third, will residents be able to see, measure and contest the results?

Local government’s comparative advantage is proximity. A resident cannot audit the Commerce Department, but she can walk into city hall, attend a council meeting and judge for herself whether the permit office got faster or the pothole got filled. AI policies at the local level should be built to leverage that proximity to increase public awareness of and support for efficacious AI initiatives. When a city deploys AI to improve service delivery, it should publish what tool is being used, for which task, and against which outcome measures (especially relative to the status quo) — then invite scrutiny. Some cities might go further and borrow from the software industry’s playbook: bug bounties that reward residents who find ways AI-assisted services fail or could improve. The key is helping everyday citizens participate in the AI rollout, rather than just be affected by it.

Contrast that with the alternative. A city that quietly embeds AI into policing, benefits determinations or code enforcement — without disclosure, without metrics, without an avenue for challenge — squanders the very trust that makes local government useful and trusted among so many. An opaque process threatens that trust, which has been acquired over decades but can disappear much faster. Every hidden or clandestine use of AI presents the risk of sowing the seeds of popular unease. The cities that instead champion transparency will likely find residents willing to tolerate experimentation. The cities that don’t will find themselves relitigating AI pilots in court and at the ballot box.

This framework will not resolve every boundary dispute between cities, states and the federal government. However, it points local officials toward the work only they can do: deploying AI in ways residents can see, judge and improve. National debates about frontier models will rage on regardless of what any city council does. Whether the permitting line gets shorter, whether the library teaches a retiree to use a chatbot, whether a laid-off game developer finds the next job — those outcomes can run through city hall.