Skip to main content
Back to reports

Artificially Inevitable

September 15th, 2026

FacebookThreadsXBluesky
Download report

INTRODUCTION 

In just a few short years, artificial intelligence (AI) has fundamentally changed the way the tech industry and the world at large operates. Driven by deep financial investments in the field, there has been a proliferation of artificial intelligence tools, models and systems, ranging in usage from chatbots to automated decision-making systems. With tech giants like Amazon, Google and OpenAI opening new offices in the city and more than 2,000 startups founded, New York City has emerged as a leading hub for AI, with the tech sector driving much of the city’s growth since the pandemic. Mayor Adams’ AI Action Plan and Governor Hochul’s Empire AI set New York City on a path to solidify its standing as one of the foremost emerging AI markets, and while it is important to acknowledge the benefits of this technology, we would be remiss to ignore the need for guardrails. 

Despite the high volume of legislation introduced around the country, enacted legislation remains limited in number and reach, with proposed legislation narrowly tailored to address specific uses and industries. Across state lines, much of the focus has been centered on consumer privacy, but some laws have also taken aim at regulating the use of AI in healthcare, journalism, entertainment, and other related industries. Local Law 144 of 2021 (commonly known as New York City’s “AI Bias Law”) was a first-in-the-nation, regulating automated employment decision tools in an effort to combat discrimination in hiring. This bill, like many others, conveys how in the absence of broad, comprehensive national legislation, states and municipalities are navigating the regulatory landscape on their own. 

With technology rapidly evolving and NYC poised to rival Silicon Valley on its current trajectory, we believe that the time to be proactive is now. With an analysis of the current regulatory landscape, this paper’s series of recommendations build on good governance practices, as well as recommendations from industry leaders. It is our hope and intent that these recommendations incite conversation, and help provide the foundational building blocks for responsible and safe use of artificial intelligence in our city. 

An Overview of AI 

To understand the conversation around artificial intelligence, one first needs to understand its history and its various functionalities. The field of artificial intelligence was first established in 1956 with the Dartmouth Summer Research Project on Artificial Intelligence serving as the foundational basis.[1 - Frana, Philip; Klein, Michael. “Encyclopedia of Artificial Intelligence: The Past, Present, and Future of AI.” Bloomsbury Publishing USA.] This summit followed Alan Turing’s 1950 landmark paper “Computing Machinery and Intelligence” wherein he devised the Turing test, the first serious proposal on machine learning. Since then, artificial intelligence has grown with forms of “traditional” artificial intelligence shaping our everyday lives from the use of search engines to virtual assistants like Siri and Alexa or algorithmic recommendation systems embedded in entertainment platforms like YouTube and Netflix. “Traditional” artificial intelligence differs from “generative” artificial intelligence, jumpstarted by the development of deep learning architecture (otherwise known as transformer architecture) in 2017, which has picked up visibility and popularity with the rapid scaling and public release of large language models (LLMs). Agentic artificial intelligence represents a new front of AI capabilities. 

Understanding Artificial Intelligence [2 - “Exploring the Shift from Traditional to Generative AI”. Massachusetts Institute of Technology, Open Learning.] 

“Traditional” AI

  • Excels at predictive analysis through pattern identification and rapid processing of data 
  • Designed to follow a set of pre-defined rules and inputs 
  • Traditional AI models are more transparent and interpretable than the black box models we see in generative AI 
  • Doesn’t require as extensive training time or substantial resources; can operate effectively with smaller datasets [3 - “Traditional AI vs. Generative AI: What’s the Difference?” University of Illinois Urbana-Champaign.]
  • Needs to be trained for each specific task or application

Generative AI

  • Learns from pattern recognition and utilizes that data to generate text, images, music, video, etc. 
  • Generative AI models often function as “black boxes” which obscures internal decision-making processes thereby making them less transparent and opaque even to system creators [4 - “What is Black Box AI?” IBM.] 
  • Requires more resources and larger datasets for output and training 
  • Does not need to be trained for each task or application; can generate content across searchable fields upon prompting 

Agentic AI [5 - “What is Agentic AI?” IBM.]

  • Has the ability to act with limited oversight for the purpose of accomplishing specific goals 
  • Models mimic human-decision making to solve problems in real time 
  • Can automate multi-step and complex tasks and procedures 
  • Can interact with external tools and the digital environment (real-time video, etc)

This distinction between different types of artificial intelligence however, while important for public understanding of the technology, fails to capture some of the more salient concerns around AI systems, in regards to risk and impact. Using the National Institute of Standards and Technology (NIST)’s Artificial Intelligence Risk Management Framework as a basis, risk is defined as the “composite measure of an event’s probability (or likelihood) of occurring and the magnitude or degree of the consequences of the corresponding event”. [6 - Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile. National Institute of Standards and Technology.] Like generative AI, traditional AI models can carry high-risk but generative AI can also exacerbate existing risks and further create new ones; thus, attempts at regulating new frontier models have largely focused on mitigating risk and impact. 

The release of ChatGPT, a generative artificial intelligence chatbot created by OpenAI in 2022, made it possible for the general public to engage with the technology with uses ranging from generating written documents or images, to translating text to a preferred language. Despite its everyday uses, the reliability of this technology has come into question with increasing rates of AI “hallucinations” — instances where models produce incorrect or misleading results that do not follow any identifiable pattern. These erroneous outputs are caused by a variety of factors such as incomplete or insufficient training data, biases from developers, and lack of contextual grounding knowledge. [7 -  “When AI Gets It Wrong: Addressing AI Hallucinations and Bias”. Massachusetts Institute of Technology.] The proliferation of AI has also sparked numerous debates from the impact of AI on education, to its potential to disrupt the labor market and creative industries, as well as the technology industry’s monumental environmental impact. 

Environmental and Economic Costs

While the environmental impact of AI has long been a point of concern, the conversations around data center needs have only become more nuanced with the first ripple effects of the AI boom beginning to take shape. Elected officials, consumers and utility companies alike are grappling with the economic and political implications of AI’s growing energy demand.

In the United States, power consumption by data centers is estimated to drive almost half the growth in electrical demand in the next 5 years. [8 - “AI is set to drive surging electricity demand from data centres while offering the potential to transform how the energy sector works”. International Energy Agency.] States with the fastest data-center growth and planned capacity include Virginia, Arizona, Nevada, Texas, Georgia and Illinois. As of 2023, energy use by data centers represented 4.4% of total U.S. electricity consumption. [9 - Shehabi, A.; Newkirk, A.; Smith, S.; Hubbard, A.; Lei, N.; Siddik, M., et al. “2024 United States Data Center Energy Usage Report.” Berkeley Lab for Energy Analysis and Environmental Impacts Division. December 2024. https://doi.org/10.71468/P1WC7Q] In areas located near data centers, an analysis of wholesale electricity prices found that electricity now costs “as much as 267% more for a single month than it did five years ago”. [10 - Saul, J.; Nicoletti, L.; Pogkas, D.; Bass, D.; Malik, N. “How AI Data Centers Are Sending Your Power Bill Soaring”. Bloomberg News. September 2025.] Given the set up of many state and regional grids, these costs have spillover effects, affecting even customers who live at a distance from centers. 

Data centers also require vast quantities of water per day for cooling purposes with the largest of centers requiring up to 5 million gallons per day, a consumption equivalent to the water use of a town with a population of 10,000 to 50,000 people. [11 - Yañez-Barnuevo, Miguel. “Data Centers and Water Consumption”. Environmental and Energy Study Institute. June 2025.] This has accelerated water scarcity in areas already under moderate to severe water stress from drought and other factors. [12 - Copley, Michael. “Data centers, backbone of the digital economy, face water scarcity and climate risk.” NPR.] 

On July 14th, 2026, New York State became the first U.S. state to impose a moratorium on the construction of large-scale data centers using 50 megawatts or more. Although far less stringent in its scope than that of the state legislature’s proposal banning data centers using 20 megawatts or more, the one-year construction ban will give New York state officials the time to create a regulatory plan addressing energy consumption and additional environmental impact. [13 - Ward, Jasper. “New York becomes the first state to impose a data center moratorium.” Reuters. (July 2026)]

Cities like Memphis, Tennessee have also seen increased air pollutants as a result of methane gas turbines powering the massive plants, sparking further concern about the disproportionate impact of these facilities in proximity to marginalized communities.[14 - “‘How come I can’t breathe?’: Musk’s data company draws a backlash in Memphis.” Politico.] As artificial intelligence continues to develop and demand accelerates, there are many in the industry who see the key role artificial intelligence plays in transforming the energy sector with AI helping shorten innovation pipelines, optimize energy systems, increase production and otherwise enhance manufacturing. [15 - “Energy and AI.” International Energy Agency.] To truly leverage artificial intelligence and further mitigate environmental impact however, deep investments are needed in renewable energy, grid electrification and other innovative green solutions. Yet these investments are moving further out of reach with every passing day, as the federal government takes steps to block and delay renewable energy projects. [16 - “Trump administration taking new steps to block wind and solar projects, undisclosed memo says.” Politico.]

AI in New York City

With the proliferation of artificial intelligence, the tech sector has emerged as a new driving force for New York City’s economy—an economy historically anchored in Wall Street and the finance industry. However, over the past decade, the tech sector has contributed to 14% of all employment growth citywide and, since the Covid-19 pandemic, it has added jobs at nearly 10 times the rate of the overall economy. [17 - Bowles, Jonathan; Dvorkin, Eli; Neches, Rachel. “Sustaining NYC’s Tech Edge”. Center for Urban Future.] As of May 2025, there were approximately 40,000 workers in NYC with AI or AI-adjacent skills. [18 - “State of the New York City Economy 2024”. New York City Economic Development Corporation.] With the launching of EmpireAI, a first-in-the-nation public-private investment consortium to advance artificial intelligence research for the public good, the region is expected to continue attracting tech talent with schools like New York University (NYU), Columbia University, the City University of New York (CUNY), the State University of New York (SUNY), Cornell Tech, the Flatiron Institute and Rensselaer Polytechnic Institute, at the helm of the initiative. 

Creating a national model for responsible AI innovation will require legislation and other policy changes that build on existing standards and frameworks, and take bold steps into the future. We lay the groundwork for this endeavor by detailing existing enacted legislation and best practice policies as informed by the city, state and federal government. 

ENACTED LEGISLATION + EXISTING POLICIES AND GUIDELINES

Federal 

In lieu of a comprehensive national law, Executive Orders have largely framed the regulatory landscape on artificial intelligence. Relevant agencies and groups such as the National Security Commission on Artificial Intelligence and the National Institute of Standards and Technology, have guided policy and provided recommendations based on their research. 

The following table lays out AI-related executive orders and memorandums from the presidential administrations of Joseph Biden (from Jan. 20, 2021 until Jan. 19, 2025) and Donald Trump (from January 20, 2025 until January 19, 2029.) 

Biden Administration

  • Executive Order 14110 (10/2023), the “Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence” 
    • Advancing a coordinated federal government approach to the development of AI, this executive order outlined eight guiding principles for agencies and departments to adhere to.[19 - Eight guiding principles for artificial intelligence: 1) it must be safe and secure; 2) it should promote responsible innovation, competition and collaboration; 3) a requisite commitment to supporting American workers; 4) policies must be committed to the advancement of civil rights and equity; 5) the interests of Americans increasingly interact with this technology must be protected; 6) Americans’ privacy and civil liberties must be protected; 7) risks for federal use of AI must be managed in order to deliver better results for the American people; 8) the federal government should lead the way in societal, economic and technological progress.]
    • Defined applicable terms; directed the development of guidelines, standards and best practices for AI safety and security; issued directives for facilitating innovation and competition; bolstered equity and fundamental civil rights, and privacy risks mitigation. 
    • Directed relevant agencies and offices to conduct research on potential harms and impact of AI on consumers and workers. 
    • Scorecard published one year after the Exec. Order. 
  • OMB Memorandum 24-10 (3/2024) - “Advancing Governance, Innovation, and Risk Management for Agency Use of Artificial Intelligence” 
  • OMB Memorandum 24-18 (9/2024) - “Advancing the Responsible Acquisition of Artificial Intelligence in Government” 
    • Established new guidance and requirements for interagency collaboration, management of AI risk and performance, and promotion of AI acquisition. 
    • Set additional practices for generative AI, utilizing frameworks and standards as published by NIST and appropriate international standards. 

Trump Administration

  • Executive Order 14179 (1/2025) - “Removing Barriers To American Leadership In Artificial Intelligence” 
    • Rescinds Executive Order 14110, the “Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence,” and directs the development of an artificial intelligence action plan. 
    • Advances a new federal policy to “sustain and enhance America’s global AI dominance.” 
    • Directs the President’s Assistant for Science and Technology (APST), the Special Advisor for AI and Crypto, the Assistant for National Security Affairs (APNSA), in coordination with the Assistant for Economic Policy, the Assistant for Domestic Policy, the Director of the Office of Management and Budget (OMB), and the heads of executive departments and agencies to develop and submit an action plan within 180 days of the executive order to achieve the aforementioned policy. 
  • OMB Memorandum 25-21 (4/2025) - “Accelerating Federal Use of AI through Innovation, Governance, and Public Trust” 
    • Rescinds Memorandum 24-10, “Advancing Governance, Innovation, and Risk Management for Agency Use of Artificial Intelligence”, and provides guidance to agencies on promoting AI innovation via a framework of providing ‘best-value’ to taxpayers. 
    • Requires agencies via CFOs to identify and remove barriers to innovation and “develop public AI strategies that elevate AI adoption and innovation as a priority.” [22- “Accelerating Federal Use of AI through Innovation, Governance, and Public Trust”. OMB.] 
    • Further requires accelerated adoption of responsible AI practices, continuing the tradition of agency-designated CAIO roles, reuse of AI code and models, and relevant reporting requirements, including agency AI use case inventories. 
    • Highlights the need to implement minimum risk management practices where ‘high-risk AI’ systems are concerned. 
    • Directs agencies to solicit public input.
  • OMB Memorandum 25-22 (4/2025) - “Driving Efficient Acquisition of Artificial Intelligence in Government” 
    • Rescinds Memorandum 24-18, “Advancing the Responsible Acquisition of Artificial Intelligence in Government” 
    • Promotes AI acquisition with cross-functional engagement, more effective risk management and tracking of AI performance, and calls for maximizing use of American-made AI.
  • Executive Order 14409 (06/2026) - “Promoting Advanced Artificial Intelligence Innovation And Security” 
    • Directs the Secretary of Homeland Security, the Director of OMB, the Assistant to the President for National Security Affairs, and the National Cyber Director to create Binding Operational Directives and other guidance prioritizing cyber defense and cybersecurity systems. 
    • Creates a timeline review period of 30 days, down from the original proposed 90 days, for new AI models. Collaboration with AI frontier developers would be on a voluntary basis. 

Additional Frameworks and Standards of Note 

  • White House National Policy Framework for AI 
    • Released in March 2026, this light-touch framework outlines six areas of interest for Congress to consider, highlighting AI’s relationship to economic development, intellectual property, consumer protection, etc. The framework also explicitly calls for the pre-emption of state attempts to regulate AI. 
  • America’s AI Action Plan
    • Released in July 2025, this Action Plan was created under the direction of Executive Order 14179. It identifies over 90 federal policy actions with the aim of enabling AI innovation and adoption, building AI infrastructure and promoting American leadership in AI diplomacy and security. 
  • AI Risk Management Framework 
    • In 2023 the U.S. Department of Commerce via NIST released a report on AI risk management, delving into AI risk and trustworthiness and how to best manage these risks.
  • National Security Commission on Artificial Intelligence (NSCAI) Final Report 
    • This 2021 report presents the NSCAI’s strategy for winning the artificial intelligence era. Each of the 16 chapters provide topline conclusions and recommendations. The accompanying Blueprints for Action outline more detailed steps that the U.S. government should take to implement the recommendations.
  • Roadmap for Artificial Intelligence Policy in the Senate 
    • This roadmap was created by a bipartisan working group in the Senate as part of the 118th Congress. Examines the impact of AI on a number of industries and issue areas, guiding relevant committees and agencies to develop legislation, mechanisms and standards. 
  • Bipartisan House Task Force Report on Artificial Intelligence 
    • This report was created by a bipartisan task force in the House of Representatives as part of the 118th Congress. Sets guiding principles and proposes recommendations and other policy initiatives the federal government should undertake across various policy areas. 
  • AI Bill of Rights 
    • In October 2022, the Biden administration put out a blueprint for an AI Bill of Rights identifying five principles to guide the design and use of automated systems: 
  1. Safe and Effective Systems; 
  2. Algorithmic Discrimination Protections; 
  3. Data Privacy; 
  4. Notice and Explanation; and
  5. Human Alternatives, Consideration and Fallback.

The federal reconciliation bill passed and signed in July 2025 previously included language to place a 10-year moratorium on enforcement of AI legislation. Negotiations in the Senate amended the House’s proposal with exemptions carved out for facilitated AI deployment and certain procedural activities. [23 - “Senate Leaders Water Down 10-Year State AI Law Ban, Instead Tie it To Federal Funding Promise.” Fisher Phillips.] The Senate proposal also included $500 million in FY2025 to carry out the “Broadband, Equity, Access, and Deployment (BEAD) Program,” awarding funding only to states in compliance with the moratorium. [24 - “The Senate Republican Budget Bill Adds Broadband Funding That Favors Musk’s Starlink and Bans State AI Laws”. Center for American Progress.] Faced with a bipartisan backlash, lawmakers eventually stripped the proposal from the final version of the bill. 

At the time of this report’s publication, new efforts to preempt state laws regulating frontier models are underway with House lawmakers introducing bipartisan legislation that would require top AI developers to create and implement plans to address risks posed by their models. Alternatively, separate attempts by the White House aim to coordinate talks between lawmakers and other stakeholders to build a legislative package around protections for kids’ internet safety and state preemption. [25 - Bordelon, B.; Haslett, C.; Miller, G. “White House’s new push to block state AI laws could ride on kids’ safety.” Politico. (June 2026)]

State

With states left to their own devices in regulating AI, New York State has not only opened its doors to innovation with the launching of EmpireAI, but in 2024, it also led the pack with the most AI-related bills. Initial first steps at regulating automated decision-making in government spaces were taken with the passage of the LOADing Act and, most recently, the RAISE Act — relating to the training and use of artificial intelligence frontier models. 

  • LOADinG Act (S7543 / A9430) - Enacts the legislative oversight of automated decision-making in government act to regulate the use of automated decision-making systems and artificial intelligence techniques by state agencies 
    • Signed into law in December 2024.
    • The legislation requires the disclosure of automated decision-making systems already in use, prohibiting the unauthorized use of automated decision-making systems by state agencies or entities on behalf of an agency and, when authorized, the law further requires the publication of an impact assessment. 
  • S3971B / A1746C - Creates a temporary state commission to study and investigate how to regulate artificial intelligence, robotics and automation; and repeals such commission. 
    • Signed into law in July 2019. 
    • This legislation establishes the commission, establishing appointment structure and further delineating roles and responsibility. 
  • Responsible AI Safety and Education (RAISE) Act (A6453B / S6953) - Relates to the training and use of artificial intelligence frontier models; defines terms; establishes remedies for violations.
    • Signed into law by Governor Hochul in December 2025. 
    • The bill would require AI companies to: implement and publish safety and security protocols before deploying frontier models; report safety incidents; implement safeguards to mitigate risk; bar the deployment of frontier models with the potential to cause “an unreasonable risk of critical harm”; and conduct third-party audits to ensure compliance with the act, imposing civil penalties up to $10 million for first time offenses and up to $30 million for subsequent violations.

Additionally, New York State is the first state in the country to require employers to disclose if AI is the reason for layoffs, a move that could help lawmakers, stakeholders and the general public understand the impact of AI on the labor market. First proposed by Governor Hochul in her 2025 State of the State Address on January 14th, the change took effect in March, amending New York State’s existing Worker Adjustment and Retraining Notification (WARN) system. [26 - “Is AI the Reason for Your Layoff? New York Becomes the First State to Require Companies to Disclose If So.” Entrepreneur.] Via the WARN system, employers are now required to fill out a form at least 90 days prior to mass layoffs or plant closure and check off whether layoffs are due to “technological innovation or automation.” [27 - New York Codes, Rules and Regulations (NYCRR) Part 921.] Employers who fail to provide notices and disclose relevant information will be subject to civil penalties up to $500 for each day the employer is in violation of the requirement.[28- Ibid.] 

Following the end of the 2026 legislative session, several bills await signature from Governor Hochul, including but not limited to: bills regulating how chatbots can interact with minors, transparency requirements for large language models, and disclosure requirements and protections for newsroom staff against AI automation. [29 - Lewis, Rebecca. “It was a pretty good year for regulating AI in New York.” City and State NY. (June 2026)]

City

In October 2023, Mayor Adams released NYC’s AI Action Plan, a roadmap outlining 37 key actions across 7 initiatives aimed at shaping the city’s policies, guidelines and standards in regards to artificial intelligence. The Plan further identifies areas where additional support and resources are required for integration of AI. As of October 2024, a year after its release, OTI was on track to have initiated and/or completed 29 of the 37 key actions. [30 - “One year ago, New York City introduced a sweeping AI plan. Here’s how it’s going.” Route Fifty.]

In January 2024, the city convened an AI Advisory Network, composed of stakeholders from across sectors, to serve on a consultative basis in supporting the city’s AI efforts. Three public listening sessions in April 2024 held by NYC’s Office of Technology and Innovation (OTI) staff revealed broad concerns about AI risks in relation to privacy, data security, fairness and bias, transparency, and human input in decision-making. [31 - “Summary of April 2024 NYC AI Public Listening Sessions.” NYC Office of Technology and Innovation.] Under the ‘Grassroots AI Education & Engagement Initiative’ (now the Artificial Intelligence Community Engagement (AICE) Initiative), first launched by Council Member Jennifer Gutierrez and then-Speaker Adrienne Adams in coordination with the GUARD Act package, over 100 listening sessions have been hosted by community organizations with the aim of engaging New Yorkers on AI and establishing a baseline literacy. [32 - “CITY COUNCIL LAUNCHES FIRST-OF-ITS-KIND GRASSROOTS AI EDUCATION INITIATIVE.” New York City Council. (November 2025)] 

Given key lessons taken from these initial forums and the foundation set forth by the initiative, we would encourage the city to plan for ongoing engagement around this issue as AI continues to evolve. The AI Public Participation and Engagement Guidance, released in December of 2025, provides tips, tools and resources for these efforts. [33 - “AI Public Participation & Engagement Guidance.” NYC Office of Technology and Innovation. (December 2025)]

Anticipating the release of more comprehensive guidelines and policies, we’ve compiled enacted legislation on the city level and other relevant guidance below. 

A recent report by NYC Comptroller Mark Levine examines five potential scenarios for AI’s impact on the job market, financial markets and the broader economy. [38 - “AI and New York City’s Fiscal Future: Planning Scenarios for an Uncertain Economy”. NYC Office of the Comptroller. May 21, 2026.] Recommendations from the report include increasing the city’s rainy day fund to absorb potential shocks to the system, protecting our pension system from market turbulence, ensuring that corporations are adopting responsible AI policies, and modernizing our own municipal systems. 

Outside of legislation pertaining to responsible and appropriate use of AI in government offices and public spaces, additional bills introduced in the New York City Council have sought to restrict the use of facial recognition in residential buildings [39 -  “Int 428-2026” New York City Council.] and public accommodation spaces like Madison Square Garden [40 - “Int 213-2026.” New York City Council.] , restrict the use of dynamic pricing in commercial businesses [41- “Int 891-2026.” New York City Council.], and impose additional reporting requirements for the use of facial recognition surveillance technology by law enforcement purposes. [42 - “Int 690-2026.” New York City Council.] Nationally, the use of facial recognition surveillance technology by law enforcement officials has become deeply divisive in the wake of brutal immigration crackdowns with documented cases of smart glasses and other facial recognition tools being used to surveil, identify and track down undocumented individuals as well as protestors. 

These encounters have sown seeds of fear and distrust in communities across the country and the use of AI-facilitated tools in these contexts has also spurred advocacy calling for the right to privacy and the dismantling of an ever-growing surveillance state. 

GUIDING PRINCIPLES & RECOMMENDATIONS

Guiding Principles

With its potential to impact numerous industries in varying ways, regulating AI raises complex policy questions and challenges. Having examined the current regulatory landscape, this report’s recommendations are grounded in the following principles, building on the city’s use guidance as well as principles enumerated in the AI Bill of Rights. 

  1. Safe and Effective Systems: Policy interventions should balance potential benefits and risks. | AI has the potential to transform countless industries, benefiting research and technology in fields such as agriculture, medicine, transportation and biotechnology. Yet in order to optimize benefits and mitigate risks, proper safeguards must also be implemented. Policymaking for AI regulation must be evidence-based and flexible, leveraging industry expertise for targeted industry-interventions.
  2. Social Responsibility: Accountability and inclusivity are essential to effective and responsible AI governance. | In order to protect consumers and spur innovation, clear standards must be established and enforced. Opacity around AI frontier models complicates informed decision-making amongst policymakers but also erodes trust in the general public. The bipartisan consensus around AI regulation offers a unique opportunity to be proactive and build a robust policy framework with community buy-in.
  3. Trust and Transparency: Early identification and ongoing management of AI risks is critical to mitigating risks. | Integrating safety standards into the early design of frontier models can prevent dangerous conditions or failures down the line. Ongoing management of identified risks through measures including but not limited to credible third-party evaluation, real-time monitoring and human intervention, can further ensure the safety of this rapid-growth technology. 

State Recommendations

With the proliferation of chatbots and other AI-powered business tools, consumers and companies alike face potential risks. This establishes the need for consumer protection, a path for legal recourse and a greater understanding of potential liabilities. 

Legislation has been introduced by various members of the NYS Senate and NYS Assembly, aiming to amend general business and civil rights laws to account for the use of artificial intelligence and its ongoing liabilities. We believe that together these bills would set the foundations in addressing these issues and further building trust in these models. 

We call on the State Legislature to pass the following bills: 

  • A222 / S5668: An act to amend the general business law, in relation to liability for false information provided by a chatbot 
  • S1169 / A08884: An act to amend the civil rights law and the executive law, in relation to the use of artificial intelligence systems 
  • S1962 / A768: An act to amend the general business law, in relation to preventing the use of artificial intelligence algorithms to discriminate against protected classes 
  • A3265: Enacts the New York artificial intelligence bill of rights to provide residents of the state with rights and protections to ensure that any system making decisions without human intervention impacting their lives do so lawfully, properly, and with meaningful oversight

We would also like to echo the need for comprehensive data privacy laws in New York State as the state currently lacks the digital infrastructure necessary for building an effective regulatory framework. My office has introduced legislation at the city level, Intro 708-2026, requiring collecting entities to obtain consent for the collection and disclosure of personal and sensitive information pertaining to New York City residents but amidst a growing data-broker economy, New York State must also set a strong foundation to more readily meet the needs of a rapidly-changing digital world and further protect New Yorkers’ right to digital privacy. 

City Recommendations

At the city level, government use of artificial intelligence should be closely monitored and reviewed. The initial rollout of the MyCity Chatbot, a NYC AI-powered chatbot designed to provide New Yorkers with information on starting and operating a business, was marred by rampant misinformation. In some instances, the chatbot falsely asserted that employers and landlords could do things long prohibited under state and federal law. [43 - Lecher, Colin. “NYC’s AI Chatbot Tells Businesses to Break the Law.” The Markup. March 2024.] While the chatbot has since been discontinued, the case should serve as a cautionary tale. As highlighted in the previous section, there is currently proposed legislation on the state level to address chatbots and other common uses, but outside of legislation, there are ways New York City can better prepare itself to meet this moment.

Procurement

In order to truly leverage AI capabilities in an appropriate and responsible manner and further address existing loopholes in backdoor data-sharing, New York City needs to reform its procurement process. Addressing what is currently flawed in the city’s procurement system and creating AI-specific standards for procurement would further enable the city to contract with suitable vendors, foster innovation across the city, and build transparent and responsible AI governance. Recognizing that effective vendor evaluation would be dependent on an individual agency’s technical capacity, we believe that technical advisement for AI-specific procurement contracts should fall under the purview of the new Office of Algorithmic Accountability, established under Local Law 188 of 2025 as part of OTI and expected to begin operations this year. 

Contractual requirements for AI procurement should: 

  • Minimize vendor lock-in and high switch costs; 
  • Outline clear metrics for vendor evaluation, and establish the roles and responsibilities of each party in ongoing testing and monitoring of models and tools; 
  • Establish early and ongoing identification and management of privacy risks; and 
  • Require data portability and Application Programming Interface (API) standardization with language emphasizing transparency. 

Contracts with the city must address potential loopholes that allow subcontractors to circumvent rules around data privacy by requiring that these regulations be adhered to in all subcontracts. Language standardization across agencies would likely cause more harm than good, as different agencies work with different constituencies, however, base guidelines for these reforms can be guided, informed and reviewed by the Office of Algorithmic Accountability. 

These recommendations reflect best practices as informed by the city, state and federal government, as well as industry leaders.

Planning for AI’s Impact on the Labor Market

Understanding the impact of artificial intelligence on the labor market will be critical to understanding how we can best support and train workers in the coming years. Emerging research has already exposed the differential impact AI is having along gendered lines with women both underrepresented in AI-augmented roles and overrepresented in sectors most at risk of disruption. [44 - Baller, Silja. “How AI is worsening workplace gender gaps and how we can course correct”. World Economic Forum.] May 15, 2025.The following bill would build off the expansion of the state’s WARN system and the recent launch of the FutureWorks Commission, established to advise the Governor on best practices for AI adoption and implementation as well as interventions to protect the economic security of workers. [45 - “Governor Hochul Launches the FutureWorks Commission to Guide Response to Impacts of AI on Workers Across New York”. NYS Office of Governor Kathy Hochul. March 19, 2026.]

  • Int 1003-2024: A Local Law to amend the administrative code of the city of New York, in relation to the creation of an artificial intelligence working group within the city commission on human rights to study its impact on employment

Regulating industry-specific uses of AI will require their own analysis and targeted intervention policies, but agencies should be guided by an overarching, unified vision. Furthermore, we must ensure that legislation regulating AI has effective oversight, enforcement and accountability mechanisms. The gaps in Local Law 144 provide a unique opportunity to go back and strengthen the legislation and further establish some credible form of evaluation. This will not only ensure that workers today are protected from bias but it will also serve as a vital tool as AI is integrated into more work places. 

In addition to internal policymaking, we believe that the city has an opportunity and an obligation to help New Yorkers better understand this technology that is rapidly changing the world we live in. With a public-facing communication strategy already built to directly engage the public on this issue, we believe public listening sessions should also go hand-in-hand with a public literacy campaign aimed at those with the least access to training opportunities and individuals and communities most likely to be impacted by AI. 

  • Public AI literacy campaign 
    • This could be facilitated through public institutions like our local libraries and CUNY campuses, as well as community-based organizations and community boards around the city. Empire AI also offers an opportunity to have some of our city’s private universities engaging with the public. 
  • Workforce training 
    • As part of the study proposed under Int. 1003, the working group should consider options for a workforce training program that targets exposed job sectors and titles. This should include training opportunities and structured pathways for workers who wish to transition to alternative careers, similar to the principle of a ‘just transition’ as commonly cited in climate justice circles. [46 - “Just Transition”. Movement Generation.]
    • Coordination with the Department of Citywide Administrative Services (DCAS) should also guide the expansion of an optional AI-skills training program for municipal workers, expanding on newly established trainings like ‘Everyday AI for Work and Life: Building Curiosity, Confidence, and Connection in the Age of Change’. [47 - “Everyday AI for Work and Life: Building Curiosity, Confidence, and Connection in the Age of Change”. NYC City Learn.] 

These are just two ways in which New York City can equip residents with a deeper understanding of how this technology impacts them and how they, in turn, can meet this moment. 

CONCLUSION

With artificial intelligence playing an ever greater role in our lives and that of our city, we believe that it is our responsibility as a public-serving office to be clear-eyed and honest about the potential this technology holds. The potential to drive innovation, advances in science and technology; to automate production, process data and information faster than any human being could, augmenting the work that we do in our daily roles and giving us back valuable time in our day. The infrastructure this technology requires is necessary for a future that moves us away from fossil fuels, but this same technology also has the potential to accelerate the devastating effects of climate change, to pump more pollutants into the air we breathe and use up one of our most precious resources. Capital expenditures on artificial intelligence data centers are driving U.S. GDP growth, but consider China’s DeepSeek model for a moment. An AI chatbot to rival ChatGPT built at a fraction of its cost, the announcement of DeepSeek wiped $600 billion off the market value for chip manufacturer Nvidia, marking the largest one-day loss for a single stock in the history of the U.S. stock market. [48 - Jamali, Lily. “It shocked the market but has China’s DeepSeek changed AI?” BBC News.] 

It is undeniable that AI will continue to evolve and impact our lives, but without guardrails and intentional policymaking, are these investments truly worth the costs, or are we simply fueling a bubble ready to pop? 

ACKNOWLEDGEMENTS

Lead Author: Cerimar Olivares, Legislative and Policy Associate

Additional support provided by: 

Rosie Mendez, Director of Legislation & Policy 

Veronica Aveis, Chief Deputy Public Advocate for Policy

Ketan Sinha, Deputy Director of Technology, Data & Development

J.C Brillant, Director of Technology, Development and Data

Kashif Hussain, Deputy Public Advocate for Infrastructure & Environmental Justice

Matthew Carlin, Deputy General Counsel

Elizabeth Guzman, General Counsel 

Dynishal Smith, Chief of Staff

Kevin Fagan, Director of Communications 

Design & Layout: Luiza Teixeira-Vesey, Digital Marketing Specialist

Cover Image: Omar Lopez Rincon, Unsplash

Thank you to our external partners: Noel Hidalgo, Executive Director at BetaNYC; 

Amen Mashariki, Director of AI and Data Strategies at the Bezos Earth Fund; 

Steven Choi, Founder and Principal at Equitas Partners.

This report can be translated into over 100 languages at no cost. Please contact us at officeadmin@advocate.nyc.gov for more information.

Our Office

David N. Dinkins Municipal Building
1 Centre Street 15th Floor North
New York, NY 10007

Sitemap
Stay Up to Date
FacebookInstagramXYouTubeLinkedInThreadsTikTokBlueskyEventbrite
Sign up for our NewsletterSign up for our Newsletter
© 2026 Copyright: Office of the New York City Public Advocate
Privacy Policy