Local governments plan to impose a sweeping regulatory framework on AI firms through city-level rules.
In the wake of a broad national backlash against artificial intelligence manifested through bans on data centers, a town in Indiana has moved to bar “artificial super intelligence” (ASI).
Recently the Bloomington City Council voted unanimously to enact an ordinance that “prohibits the distribution and deployment” of ASI. Backed by the Safe AI Forever Institute, the ordinance defines ASI as a “system or entity capable of significantly outperforming humans on essentially all cognitive tasks” and able to operate without “continuous human oversight.” Under the measure, any model within Bloomington’s borders cannot be activated via any user interface, regardless of where the underlying developer is located. Nearby Indiana University students won’t be kicked off ChatGPT immediately, since the city has not yet designated any models as “superintelligent.”
Councilmember Dave Rollo introduced the measure in a memo to the Common Council, the legislative body of the City Council. He cited concerns from AI experts and developers about the technology’s potential to cause “catastrophic outcomes” for humanity, including human extinction. “The level of risk is simply unacceptable,” wrote Rollo, citing “many AI experts” who place the probability of extinction between 50 percent and more than 99 percent.
Rollo described the policy as an “interim public safety parameter…during a period of state and federal legislative inaction,” but the requirements for lifting the ban reveal a more complicated intention. According to Rollo’s memo, the ban would last until Indiana enacted a “comprehensive framework for Provably Safe Superintelligence,” after which it would expire after 91 days. Otherwise, it would remain in effect until the council determines a “broad scientific consensus that the development of Superintelligence shall be done safely and controllably.” For the council, that consensus would need to be demonstrated by a probabilistic risk assessment that shows 99.9999 percent certainty of safety—a standard typically applied to nuclear power plants.
“This is an impossible standard to meet,” said Adam Thierer, a senior fellow at the R Street Institute, to Reason. “There’s no way to scientifically measure that. Second of all, there’s nothing in this world that’s 99.999 percent safe. Nothing.”
“It’s likely that this would stay in place pretty much forever once it’s on the books,” Thierer added. When the ordinance was introduced, it would have allowed the ban to be lifted if 95 percent of the city’s residents voted to overturn it; that condition was removed before the council’s vote.
Other cities, counties, and states have adopted their own restrictions on specific AI-related technologies, including automatic license plate readers, AI pricing, and data centers. What sets AI models apart from license plate readers or data centers is their ability to cross state lines instantly, all the time. Consequently, developers cannot easily tailor a large language model to fit a single jurisdiction, and local policies can influence models far beyond their borders.
That, it seems, is the aim behind Bloomington’s ordinance. The measure, Rollo writes, “could set a precedent and a template for other cities to follow. Over 100 million people live in U.S. cities of Bloomington’s size or larger. If roughly 350 such cities enacted similar ordinances in the coming weeks or months, this would impose a massive regulatory burden on firms developing ASI.” The memo also notes that the Safe AI Forever Institute and other nonprofit groups will continue lobbying for similar restrictions in other municipalities.
Their strategy, then, is to fashion a regulatory regime through municipal policies that would effectively govern frontier AI models across the country. California and New York have already employed a similar approach in some form.
A patchwork of state policies is already confusing, but layering hundreds of municipal rules would create “a compliance hell,” Thierer tells Reason. “And I think that most of the sponsors not only realize that, but now understand that’s the point.”
The ordinance could be subject to legal challenges. The Dormant Commerce Clause restricts state laws that unduly impede interstate commerce, even in the absence of federal law. If states lack the power to regulate interstate commerce, then counties clearly shouldn’t have that power either. Thierer also warns that the measure could invite First Amendment lawsuits for restricting people’s “ability to learn about these technologies or utilize them.” No lawsuit has been filed yet.
The ban could hit Indiana University particularly hard, home to the Luddy Artificial Intelligence Center, which includes hundreds of researchers and students devoted to studying AI. Beyond that, AI fluency is rapidly becoming essential for delivering value in any field, a reality now subject to Bloomington’s city governance.
It appears these policies may not end with Bloomington. “It could be that we’re witnessing the emergence of a new faction within the AI safety movement—centered in the Safe AI Forever Institute and allied groups—aimed at creating a broader vetocracy at the municipal level,” Thierer tells Reason. “Just as many people have shown up at town halls, county planning meetings, and zoning hearings to protest data centers, you could see similar mobilization now aimed at AI models.”
