DomainHuntdomainhuntAdvertise

Bloomington Mayor Vetoes Superintelligence Ban, Citing Enforcement and Privacy Risks

Bloomington Mayor Kerry Thomson has vetoed a proposed ban on distributing and deploying artificial superintelligence, warning that unclear legal authority, enforcement challenges and potential digital surveillance could threaten research and innovation. Sponsor Dave Rollo is open to negotiating a revised ordinance.

GovernmentThe B Square Bulletin ·

Bloomington Mayor Kerry Thomson has blocked a proposed municipal ban on artificial superintelligence, putting a local effort to restrict the development and distribution of potentially dangerous AI systems at a crossroads. Her decision highlights a difficult question for governments considering restrictions on advanced AI: how can they regulate technology that may operate remotely, cross jurisdictional boundaries and be difficult to identify without examining private digital activity?

In a statement issued on November 19, Thomson declined to sign an ordinance that Bloomington's city council had approved unanimously among members present earlier in the month. She acknowledged the risks associated with increasingly autonomous artificial intelligence but argued that the proposed legislation could undermine the city's research economy while creating legal and practical problems for officials responsible for enforcing it. The measure's sponsor, Councilmember Dave Rollo, has indicated that he would rather explore revisions with the mayor than immediately pursue an override.

The veto puts AI policy at a crossroads

The proposed ordinance would establish a local prohibition on distributing and deploying artificial superintelligence within Bloomington, Indiana. It represents a more direct intervention than the city's earlier nonbinding statements about advanced AI risks, seeking to make certain activities violations of municipal law rather than simply calling on other governments to act.

Thomson's decision does not amount to a rejection of the underlying concern about increasingly capable AI. Instead, her objection centers on whether the city can responsibly implement the particular restrictions Rollo has proposed. She pointed to the potential consequences for Bloomington's innovation economy, unresolved questions about the municipality's regulatory powers and the practical demands of identifying violations.

The mayor also emphasized the distinction between supporting research and permitting every possible direction of AI development. Artificial intelligence already contributes to work in medicine, education and research, she said, and the city wants to attract people and organizations working in those fields. She warned that legislation with unclear boundaries could discourage companies and researchers from investing locally.

According to The B Square Bulletin, the mayor's statement came after several amendments had been added to the proposal. Those changes were intended to narrow its scope and address concerns raised after Rollo first introduced the measure. Despite those revisions, Thomson concluded that important questions remained unanswered.

Her position creates a policy dilemma for the council. Bloomington has expressed concern about the dangers of advanced AI, but converting that concern into an enforceable local restriction requires definitions, legal authority and procedures that the ordinance does not fully establish.

Rollo wants action on ASI

Rollo's campaign is driven by his belief that artificial superintelligence presents a threat that governments cannot afford to ignore. He describes the technology as potentially dangerous on a scale that affects humanity broadly, rather than merely presenting a conventional regulatory challenge for software companies.

Speaking to The B Square Bulletin by telephone after the veto announcement, Rollo said he remained hopeful that he could work with Thomson on a version of the legislation she would accept. He characterized superintelligent AI as a deadly risk and argued that technological development is moving toward the relevant capabilities quickly.

His concern is not limited to the possibility that AI systems could become more capable than current applications. He is particularly focused on systems that could operate autonomously, perform intellectual work across a wide range of disciplines and improve their own underlying technology without continuous human direction.

Rollo argues that the leading companies developing AI are pursuing capabilities that could eventually meet that description. Although he acknowledges that artificial superintelligence has not yet emerged, he believes the possibility of its arrival is sufficiently urgent to justify preventive action.

At the November 18 council meeting, he presented his argument partly by distinguishing the technology covered by the ordinance from the AI applications already being developed and used in Bloomington. He maintained that local businesses and Indiana University should not be treated as though their existing work were the target of the proposal.

That distinction is central to his effort to reassure the city's research community. Rollo wants the ordinance to address a particular category of future autonomous systems, not interfere with ordinary AI development or responsible research. However, the mayor's concerns show that the distinction must also be reflected in language that officials can apply consistently.

What counts as superintelligence

The ordinance attempts to define artificial superintelligence through a combination of characteristics involving autonomy, general capability and intellectual performance. It describes systems able to change their own algorithms, execute computer code beyond restricted environments and complete complex activities without ongoing human supervision.

Those operational characteristics are combined with a broader requirement: the systems would need to possess cognitive capabilities substantially beyond those of human beings. The intended category is therefore much narrower than the everyday AI products that residents and companies currently encounter.

This distinction matters because contemporary generative AI tools perform recognizable tasks, including producing text, writing computer code and generating images. Their capabilities can be impressive, but that does not automatically make them artificial superintelligence under the ordinance's proposed definition.

Rollo contrasted ASI with the human-centered AI work taking place at Indiana University. He described that research approach as emphasizing ethical applications, human oversight and positive social outcomes. He also praised local technology companies and university researchers for their approach to AI development.

However, the ordinance does not establish a formal evaluation process for determining whether a particular system meets its definition. It also does not identify an independent technical organization that would make that determination.

This leaves a significant implementation problem. Officials would need to establish whether a system's capabilities satisfy the ordinance's criteria, yet the proposed legislation does not provide a standardized test or an independent certification mechanism. Different interpretations could therefore complicate decisions about which products and companies fall within the prohibition.

For a technology category that has not yet materialized, defining the boundary in advance is particularly difficult. The legislation attempts to anticipate a future capability, but the city would still need a practical method for recognizing it if and when it emerges.

Digital distribution crosses borders

The geographical reach of the proposal creates another challenge. Its restrictions would apply to distribution and deployment within Bloomington, including delivery through websites, software and other electronic interfaces. The physical location of the computers hosting an AI system would not necessarily determine whether the prohibition applied.

Consequently, an organization based outside Indiana could potentially fall within the ordinance's reach if it made a prohibited system available to someone inside Bloomington. The proposal seeks to regulate the availability of a digital service within the municipality, rather than limiting its attention to businesses physically operating there.

That approach raises questions about how far local government authority extends across digital networks. A municipality normally exercises its regulatory powers within defined geographical boundaries. Online services, however, can be delivered across those boundaries without the provider establishing a physical presence in every location where someone accesses them.

The ordinance invokes Indiana's Home Rule Act as a basis for municipal action. Home rule generally gives local governments authority to address local affairs within the limits established by state law. Rollo's proposal compares superintelligence with hazardous materials and industrial activities that municipalities may regulate to protect public safety.

Thomson's statement specifically questioned whether Bloomington has the legal authority required to regulate the technology in the way the ordinance envisions. The legislation does not resolve whether the city's powers can support a prohibition affecting companies and digital systems operating beyond its borders.

The proposal would authorize Bloomington's legal and police departments to pursue civil penalties and court orders against entities violating the prohibition. Those enforcement options would be discretionary, meaning officials would not be automatically required to pursue every possible case.

But authorization to take legal action does not answer how the city would discover violations. Establishing that a prohibited system is being distributed locally could require information about online services, their users and the activities taking place through digital interfaces.

Privacy concerns complicate enforcement

Thomson's most pointed objection concerns the possibility that enforcing the ordinance could bring the city into conflict with residents' digital privacy. Determining whether a prohibited system is being accessed or distributed might create pressure for officials to examine private online activity.

She connected that possibility to Bloomington's recent efforts to restrict government surveillance technologies. Only five weeks earlier, on October 14, the city council had voted unanimously among members present to prohibit automated license plate readers and certain surveillance camera systems that permit outside organizations to access recorded information.

The mayor argued that a new law should not create an enforcement obligation that conflicts with those privacy protections. Although the superintelligence ordinance does not explicitly establish additional internet surveillance powers, Thomson is concerned about what investigators might need to do to establish violations.

Her objection is therefore about the practical consequences of enforcement, rather than a claim that the ordinance itself directly authorizes monitoring residents' internet communications. If officials cannot identify prohibited distribution without examining private activity, the city could face a difficult choice between enforcing the law and preserving the privacy principles it has recently strengthened.

Rollo disputes that interpretation. He told The B Square Bulletin that the city would not need to surveil residents to investigate possible violations. Instead, he suggested two approaches: responding to complaints submitted by members of the public and testing whether a prohibited system could be accessed using a computer operated by the municipality.

Such testing could establish whether a service is available from a city location without necessarily examining an individual's private communications. However, the source does not report that a detailed enforcement protocol has been adopted, and the ordinance itself leaves important operational questions open.

The disagreement illustrates the difference between identifying a possible enforcement method and establishing a complete system that protects privacy while producing legally sufficient evidence. Even if complaints and municipal testing could identify some violations, the legislation does not specify how officials would handle every possible distribution scenario.

"We don't need to surveil anyone."

— Dave Rollo, Bloomington city councilmember

Four amendments narrowed the ban

The council adopted four amendments before approving the ordinance on November 4. Rollo himself introduced all four, seeking to address problems identified in the original proposal and reduce the possibility that the legislation would interfere with ordinary AI users or legitimate research.

One amendment explicitly removed end-user access from the prohibition. Under the revised approach, the ordinance would target organizations distributing or deploying superintelligence, rather than people who simply use an AI service made available to them. A Bloomington resident would therefore not violate the proposed rule merely by accessing a system through a website or another electronic interface.

Another revision removed an unusually demanding requirement that would have governed repeal of the prohibition. Rollo's original proposal would have required approval from more than 95% of registered voters, together with scientific findings establishing an exceptionally high level of certainty about the technology's safety, before the ban could be withdrawn.

The council removed those conditions from the operative provisions. A scientific-safety objective remained in the ordinance's introductory language, but the extraordinary referendum and scientific-certainty requirements would no longer determine whether the city could repeal the restriction.

The remaining two amendments addressed terminology and policy context. One revised the language to distinguish artificial superintelligence from artificial general intelligence. The other added a reference to the council's previous resolution addressing advanced AI risks.

Artificial general intelligence, or AGI, is a term generally used for AI capable of handling a broad range of intellectual tasks rather than being limited to a narrow application. ASI describes a hypothetical further level of capability that would exceed human cognitive performance across virtually all fields. The distinction is important because the two terms describe different technological thresholds.

The amendments reduced the proposal's potential reach, particularly for ordinary users and local technology activity. Yet they did not settle the fundamental questions about technical classification, interstate digital distribution or the city's ability to investigate suspected violations without intruding on private activity.

That is why Thomson's objections persisted even after the council revised the measure. The changes addressed particular concerns about wording and scope, but the mayor believes the city still lacks sufficiently clear answers about how the prohibition would operate.

Bloomington's earlier AI efforts

The veto is the latest development in a campaign by Rollo to persuade Bloomington's government to address what he regards as an existential threat from increasingly powerful artificial intelligence. The current ordinance is his third attempt to bring the issue into the city's formal policymaking process.

His first effort came in July 2023, when he proposed a resolution calling for a temporary halt to artificial general intelligence development until adequate safeguards could be established. He subsequently withdrew that proposal after Indiana University officials raised concerns about possible consequences for research funding and support among fellow councilmembers appeared insufficient.

On June 10 of this year, the council took a different approach. Members unanimously adopted a resolution calling on higher levels of government to address existential AI risks. It also urged a pause in developing advanced AI until the technology could be aligned with human values and well-being.

That earlier measure expressed the council's position but did not establish an enforceable local ban. It was directed toward governments with broader regulatory powers and did not make distribution or deployment of superintelligence a violation of Bloomington's municipal code.

The November ordinance represented a substantial change in approach. Instead of relying exclusively on appeals to state, federal and international authorities, it attempted to establish a direct local restriction. That shift is central to the disagreement now facing the mayor and council.

Rollo sees municipal action as a way to create pressure from the ground up. His broader strategy envisions hundreds of cities adopting comparable restrictions, building a wider political movement that could eventually encourage more comprehensive rules at state, national and international levels.

He told councilmembers that AI-safety advocate Nate Soares contacted him after the November 4 vote. According to Rollo, Soares congratulated Bloomington and expressed hope that other municipalities would follow its example. Rollo sees such developments as a possible foundation for broader restrictions on artificial superintelligence development.

He also cited new research and warnings from AI experts during the November 18 meeting. His argument was that technological progress is accelerating and that opportunities to prevent the emergence of dangerous systems could narrow as development continues.

The mayor and the ordinance's sponsor therefore approach the same broad concern from different institutional perspectives. Rollo wants local government to act before the technology arrives. Thomson wants a proposal that can be implemented within the city's legal powers without damaging research activity or compromising residents' privacy.

Six votes needed for override

The immediate consequence of the veto is a decision for Bloomington's nine-member city council. Under Indiana law, the council can attempt to restore a vetoed ordinance by approving it again with a two-thirds majority at its first regular or special meeting after the mayor's decision period.

The city clerk's office recorded that the ordinance was delivered to Thomson on November 9 at 5:13 p.m. Her deadline to sign it or return it with a veto message was November 19 at the same time. Under the applicable state procedure, failing to take either action would also have resulted in a veto.

The council's next regular meeting is scheduled for December 2 at 6:30 p.m. An earlier special meeting could provide another opportunity, but the scheduled December meeting is the immediate occasion identified for considering an override.

Nov 4Council approves the ordinance 7–0
Nov 9Ordinance delivered to the mayor at 5:13 p.m.
Nov 19Deadline and veto announcement
Dec 2Scheduled council meeting at 6:30 p.m.

Six affirmative votes would be necessary, regardless of how many members attend. The November 4 vote was 7–0, although two councilmembers, Sydney Zulich and Kate Rosenbarger, were absent. That result demonstrated support among the members present, but it does not guarantee that six members will favor overriding the mayor.

Councilmembers could conclude that Thomson's objections justify additional negotiations or changes to the measure. They could also decide that the ordinance's broader objective is important enough to warrant another vote in its current form. The outcome depends on the council's response to the unresolved legal and implementation issues.

Bloomington has recently experienced a similar disagreement over municipal legislation. In June, the council approved an ordinance requiring seasonal vehicle closures on Kirkwood Avenue by a 5–4 vote. Thomson vetoed it, citing issues involving public participation, costs and implementation.

When the council considered overriding that veto in July, the effort failed. The vote was 4–4, with Rollo among those who did not support overriding the mayor despite having voted for the original ordinance. The episode demonstrates that an initial vote in favor of a proposal does not necessarily translate into support for overriding a mayoral veto.

A possible path to revision

Thomson's statement left the door open to a different version of the superintelligence legislation. She said she had discussed the proposal extensively with Rollo and appreciated his willingness to consider changes. Her stated preference is to continue discussions aimed at protecting Bloomington while preserving the city's research and innovation economy.

The mayor made clear that she would not approve the existing text. However, her comments indicate that she is not necessarily opposed to all municipal safeguards addressing advanced AI. The distinction leaves room for negotiations over clearer definitions, practical enforcement mechanisms and the limits of the city's authority.

Rollo interpreted the statement as an opportunity for collaboration. After speaking with The B Square Bulletin, he said he was inclined to pursue that route rather than immediately organize an attempt to override the veto.

He argued that working with the mayor could produce a stronger outcome than treating the disagreement as a confrontation between the executive and legislative branches of city government. His preference suggests that the immediate political contest may give way to discussions about the ordinance's technical and legal provisions.

Any revised proposal would need to address the problems Thomson identified. That includes establishing how officials could determine whether a system meets the definition of ASI, how a local prohibition could apply to remotely hosted services and what evidence could support enforcement without requiring intrusive monitoring of residents.

The city's research community also remains an important consideration. Rollo's distinction between hypothetical superintelligence and existing human-centered AI applications is intended to reassure researchers and businesses. A revised measure would need language that makes those boundaries sufficiently clear to address the mayor's concern about discouraging investment.

Why it matters

Bloomington's dispute illustrates a growing challenge for policymakers attempting to address risks from hypothetical future AI systems before those systems exist. Legislators can agree that increasingly autonomous technology deserves serious attention while disagreeing over the definitions, legal powers and enforcement arrangements necessary to regulate it.

The superintelligence question also differs from conventional software regulation. The proposed Bloomington restrictions would reach digital services delivered through the internet, potentially involving companies and computer infrastructure located outside the municipality. Establishing what technology qualifies as ASI adds another layer of difficulty because the ordinance does not provide a formal testing or certification process.

The privacy issue is equally significant. A law designed to prevent access to a potentially dangerous technology could create pressure for authorities to investigate online activity. Thomson's concern is that a municipality should not adopt a restriction without understanding what its enforcement would require, particularly after the council's recent action against certain surveillance systems.

For supporters of stronger AI safeguards, Rollo's proposal represents an attempt to make local government part of a wider movement for preventive action. Its potential influence, in his view, would come not only from what Bloomington could enforce but also from whether other municipalities adopted similar policies and encouraged broader government intervention.

For researchers and technology companies, the outcome could clarify how far Bloomington intends to go in distinguishing speculative superintelligence from current AI development. The mayor has explicitly connected that distinction to the city's ability to attract investment and sustain its innovation economy.

The next development to watch is whether the council pursues an override on December 2 or prioritizes negotiations over revised language. Rollo has expressed interest in working with Thomson, while the mayor has signaled that a different proposal could still receive consideration. The central question is no longer simply whether Bloomington recognizes the risks of superintelligence, but whether its officials can design a legally defensible and enforceable response without undermining the research and privacy protections they also want to preserve.

Read the original at The B Square Bulletin

More news

Written by DomainHunt from reporting by The B Square Bulletin. Read the original article for the full story. Not financial advice.