New York clamps down on AI applications locally after Washington steps back in favor of corporate self-regulation
In the Webster Hall auditorium in Manhattan, New York Mayor Zahran Mamdani chose to dismantle the most promoted narrative in Silicon Valley: the idea that the sweeping expansion of artificial-intelligence technologies is an inevitable destiny from which there is no escape. In front of a gathering organized by the AI Now Institute, Mamdani, accompanied by activist Ilana Glazer, delivered a direct message to the city’s residents urging them to reclaim the initiative and reject surrendering to fears of job loss or the hijacking of children’s minds, at a moment when the federal administration in Washington opted for an almost complete retreat from enacting binding legislation, leaving cities and municipalities to face tech giants directly.
The federal retreat from oversight in favor of self-regulation opens the door for companies to evade accountability.Mamdani criticized the reduction of the national debate in the United States to a “arms race” narrative with China, a framing that is being invoked to push humanitarian and social considerations aside in the name of national competition. This trajectory was reinforced by the announcement earlier this month of the “Superintelligence Task Force” attached to President Donald Trump, which focused its efforts on enabling developer companies to self-regulate rather than imposing strict legal constraints, despite the escalation of internal warnings issued by employees who disclosed transgressions within Anthropic, Google DeepMind and OpenAI concerning risks that could threaten human existence.
Mamdani linked this legislative retreat to direct financial interests whose payouts are approaching on Wall Street. According to his statements, the attempt to portray artificial intelligence as an infallible system that cannot be questioned and that only its creators can regulate revives the same arguments that Facebook promoted in 2008 to evade accountability and protect company valuations before their shares went public. This confrontation comes at a sensitive moment, as Anthropic is expected to go public this year, while OpenAI has set 2027 as the date for its anticipated IPO, thereby doubling the incentives to conceal flaws and sidestep regulatory constraints in order to maximize market value.
New York City moved with practical measures to fill the legislative gap and cement oversight.Mamdani issued a decree last month banning the use of artificial-intelligence tools in elementary and middle schools after concerns grew about their forced deployment in classrooms without parental consent. He reinforced this with a policy approved during the summer that requires property owners, brokers and real-estate platforms to provide explicit disclosure when using AI to edit property listing photos or videos, in order to protect tenants and buyers from visual deception and fraud.
Mamdani warned against reckless adoption of the technology without recognizing its structural limitations, citing officials in U.S. Department of Defense investigations reported by Bloomberg about an American military strike that hit a school in Iran in February, killing more than 150 people, including at least 123 children, where it was found that over-reliance on targeting algorithms was one of the contributing factors. These facts reveal the steep price of portraying computational models as infallible tools and delegating life-critical decisions to them without precise human oversight.
Regional reliance on providers’ voluntary commitments entails a direct strategic risk.This municipal clash in New York maps out a transformation that affects institutions in the Gulf, Egypt and the Arab Levant; as banks, government agencies and real-estate development firms in the region accelerate the integration of these generative models into their services, betting on “self-regulation” by American tech giants becomes a step with uncertain consequences. Handing over models safely to companies merely on the basis of providers’ assurances exposes local automation projects to risks of real-estate and financial misinformation or unexpected algorithmic flaws. What changes today for you as a technology or executive leader in Riyadh, Abu Dhabi or Cairo is the need to shift from being a consumer of ready-made cloud services to building strict local audit and testing frameworks that impose mandatory disclosure standards on AI outputs directed to the public, and that prevent delegating vital decisions to reasoning systems whose developer companies evade legal responsibility before they go public.
Mamdani’s call to build a different framework for AI accountability reveals the reality of the next global stage: software marketed as infallible systems remains riddled with fundamental limitations, and relying on it without strict institutional oversight puts users in direct confrontation with errors whose developers do not assume responsibility.