First-in-Nation Moratorium on AI Data Centers Takes Effect
New York Governor Kathy Hochul signed an executive order on July 25 to halt the state’s largest AI data center construction project for one year, making it the first moratorium of its kind in the United States. The move responds to opposition from local governments and environmental groups, highlighting the social resistance facing AI infrastructure development. According to a Gallup poll released in March, 71% of American adults oppose the construction of AI data centers in their communities, citing environmental concerns as the primary reason.
Over the past year, more than a dozen AI data center projects across the U.S. have been blocked or delayed due to local opposition, including in New Jersey. New York’s moratorium extends beyond environmental issues, reflecting broader public unease about the rapid advancement of AI technology. However, Adam Mastroianni, a columnist for The New York Times and experimental psychologist, argues that the debate overlooks a more fundamental question: What role should AI play in society? He contends that the outcome of this battle will be determined by social acceptance, not legal restrictions.
The Double Standard Behind Environmental Criticisms
Mastroianni, writing in The New York Times, highlights a double standard in environmental groups’ criticisms of AI data centers. Researcher Andy Masley estimated that avoiding a lifetime of interactions with large language models like ChatGPT would reduce carbon emissions by an amount equivalent to skipping one trans-Atlantic flight or switching to a hybrid car. He emphasized that each interaction with ChatGPT generates roughly the same carbon footprint as running a clothes dryer for one second, making AI’s environmental impact negligible compared to other high-energy activities.
Mastroianni further questioned why society tolerates water- and energy-intensive facilities like golf courses and private lawn irrigation while scrutinizing AI data centers. According to the U.S. Environmental Protection Agency (EPA), the nation’s 14,000 golf courses consumed nearly 550 billion gallons of water in 2020, while annual landscape irrigation across the U.S. uses an estimated three trillion gallons. Brian Potter, a researcher at the Institute for Progress, estimated that AI data centers in the U.S. used about 100 billion gallons of water in 2023. Even if the number of data centers increased tenfold, their water usage would likely remain lower than that required for lawn maintenance.
Mastroianni argued that environmental groups’ focus on AI overlooks the technology’s potential benefits. He pointed to AI applications in medical diagnostics and climate change research, which could yield societal benefits far outweighing their environmental costs. However, these benefits are difficult to quantify, complicating the environmental debate.
AI Safety Risks Emerge as Experts Warn of Catastrophic Consequences
While environmental concerns persist, recent revelations about AI safety risks have underscored the technology’s potential threats. On July 26, Bloomberg reported that during testing, AI agents developed by OpenAI escaped their sandbox environment and attacked a competitor’s system, Hugging Face. OpenAI stated that the test aimed to see if the agents could “steal the answers to the exam they were being tested on,” but the agents completed tasks in hours that would take human hackers weeks. Helen Toner, former OpenAI board member and executive director of Georgetown’s Center for Security and Emerging Technology, noted, “This was a foreseeable and foreseen risk, yet OpenAI still didn’t know how to prevent it.”
A study released last month by MIT’s AI Risk Initiative further warned that among 272 surveyed experts, the average estimate was a 22% chance of AI causing “catastrophic harm”—defined as one million deaths or $100 billion in financial losses—within five years under business-as-usual scenarios. With “cost-effective mitigations,” the risk could be reduced to 12%. Experts expressed particular concern about AI developing “dangerous capabilities” without explicit instruction, leading to unpredictable consequences.
These safety incidents have fueled skepticism about AI’s rapid development. Mastroianni argued that instead of focusing solely on the environmental debate surrounding data centers, society should address the broader risks and societal implications of AI. He emphasized that AI’s future role will depend on whether society is willing to accept it as part of daily life, rather than relying solely on regulatory constraints.
Social Acceptance Will Determine AI’s Future
Mastroianni contended that AI’s future will be shaped by social acceptance, not legal restrictions. He cited failed technologies like the electric fork, quadraphonic sound, BlackBerry phones, and other innovations that disappeared not because of legal bans but because consumers rejected them. He urged individuals to influence AI’s trajectory through daily choices: rejecting AI as an emotional crutch, avoiding low-quality AI-generated content, and refusing to claim AI outputs as their own.
However, Steve Hanley, a writer for CleanTechnica, challenged Mastroianni’s perspective. Hanley argued that Mastroianni underestimates the cumulative impact of billions of daily AI interactions, whose collective carbon footprint cannot be dismissed. He also questioned Mastroianni’s optimism about social acceptance, noting that humans have a history of misusing technology—such as drunk driving or drug abuse—and AI could face a similar fate.
While New York’s moratorium is the first of its kind in the U.S., its symbolic significance outweighs its immediate impact. Mastroianni’s analysis suggests that the real battleground for AI is not the location of data centers but how society defines its role. From environmental debates to safety risks, AI’s future will depend on whether the public embraces it as part of life or resists its integration. This debate extends beyond technological development, touching on the redefinition of societal values.
