Bloomington, Indiana Bans Superintelligence: Ordinance 2026-29 Explained

On November 4, 2026, Bloomington, Indiana’s Common Council voted unanimously to prohibit superintelligent artificial intelligence inside city limits — not only in a data center on Main Street, but also when a resident opens a browser or API session to a model hosted anywhere else. The Indiana Daily Student reported the vote during a long council night that also covered development code and bonds. The measure is Ordinance 2026-29: Chapter 14.25 of the Bloomington Municipal Code, pitched as an interim public-safety perimeter until Indiana or the federal government acts.

TL;DR: what changed on November 4

Primary sources: council memo versus newsroom summary

The staff memorandum and ordinance text published on bloomdocs.org are the authoritative wording. Councilmember Dave Rollo and the council office frame the rush toward artificial superintelligence (ASI) as an acute emergency while state and federal governments have not set safety rules. The memo cites whistleblower resignations, model containment failures reported in the press, and Resolution 2026-10 (June 2026) urging officials to study existential AI risk — the political runway that preceded this ordinance.

IDS paraphrases the definition: AI that outperforms humans at nearly all mental tasks, can act autonomously, may rewrite its own code, or run without continuous human oversight on tasks touching infrastructure, markets, or public safety. That is journalistic shorthand; builders should quote 14.25.x sections in the PDF when arguing compliance.

SAFE AI Forever Institute’s post-vote summary aligns on the 7–0 outcome and Chapter 14.25 title, but explainx.ai treats the city PDF and IDS as the core legal record, not advocacy group press releases.

What the ordinance tries to do — in plain language

Gavel on a scroll with a green band, for Bloomington Ordinance 2026-29 superintelligence municipal law

Ban the scary endpoint, not all ML. Rollo’s memo distinguishes narrow or tool AI from autonomous superintelligence that could recursively self-improve beyond human control. That matches how federal superintelligence discourse and Sanders–Casar legislation talk — but here the enforcer is a municipal peace-and-safety chapter, not Congress.

Reach through the internet. Section 14.25.030 treats deployment as remote execution through a local interface inside Bloomington. B Square Bulletin’s analysis walked through the consequence before the vote: a Californian model accessed from a Bloomington laptop could plausibly fall under both user and distributor liability, though the text does not allocate responsibilities line by line.

Buy time, not rewrite physics. The ordinance declares itself interim — it yields to Indiana if the General Assembly passes a comprehensive framework, then dies 91 days later. Until then, Bloomington asserts home-rule police power under Indiana Code § 36-1-3 to treat uncontrolled ASI as a public nuisance, analogizing to local hazardous-material bans in the memo.

Make lifting the ban nearly impossible. Without state law, dissolution requires council findings of 99.9999% (“six nines”) catastrophic-risk certainty — language borrowed from nuclear plant engineering culture in the memo — and approval from more than 95% of registered voters for superintelligence development. B Square flagged that Indiana may not authorize a city to run such a referendum without statute, and that a future council could still repeal the ordinance through normal legislative process despite the high bar to lift the substantive ban.

Enforcement, fines, and who actually gets targeted

IDS says the city may seek civil penalties, injunctive relief, and Monroe County prosecutor referrals. B Square’s October analysis pegged continuing violations at up to $2,500 per day under Bloomington’s general penalty schedule, while stressing enforcement discretion — Bloomington is not required to chase every violation, and the fiscal impact statement predicts no new city spending.

Practically, that means:

1. No immediate police raid on GPUs — unless a future complaint and city law department choose to act.
2. High ambiguity for frontier labs — the ordinance does not name OpenAI, Anthropic, Google, or xAI; it names a capability class without a test.
3. Leverage for civil litigants — Rollo’s memo argues local bans help plaintiffs meet burden-of-proof arguments in imminent-harm lawsuits, echoing Treasury Secretary Scott Bessent’s reported stance that Washington might not blanket-shield frontier developers.

For AI builders, the operational question is not “is GPT-6.1 illegal tomorrow?” It is whether any system marketed as ASI or described in marketing as superhuman across domains creates municipal exposure when a Bloomington employee uses it at work.

How Bloomington fits the wider superintelligence map

Bloomington is a college town of roughly 80,000, not a federal regulator. Its move matters because it is among the first city codes that ban a capability class by definition in the superintelligence bucket rather than regulating deepfakes, government procurement, or data centers — topics explainx.ai covered in New York Council validation bills, Pennsylvania deceptive-AI advertising law, and California data-center moratoriums.

It also sits on the coalition politics axis: Bengio, Hinton, and unlikely allies backing federal superintelligence bans versus industry pushback that the term is undefined. Bloomington codifies one local definition and dares courts to say whether home rule reaches packets crossing state lines.

Rollo’s memo explicitly hopes ~350 similar cities copy the template, working with groups like SAFE AI Forever to stack municipal friction on frontier labs. Even if copycat ordinances fail legally, procurement committees and university IRBs in university towns may treat the vote as a signal for research contracts.

What developers and labs should do now

Map “deployment” literally. If your product is a browser tab, an API key, or a desktop agent, Bloomington’s text cares where the human sits, not where the weights live. Document geo policies in terms customers understand — the same way teams already handle export controls on model weights.

Do not self-certify as superintelligent in marketing. Ordinance fights will land on labels and capability claims, not benchmark tables. Posts debating whether Astra-class models count as superintelligence are fun academically; municipal code cares about rhetoric and autonomy features like unsupervised self-improvement loops.

Separate research from production bans. Indiana University reportedly worried about research funding when Rollo floated an earlier 2023 resolution, per B Square’s history section. Teams doing campus R&D should watch IU policy and city enforcement statements after mayoral sign-off — not just federal Anthropic safety lobbying.

Watch state preemption. If Indiana passes provably safe superintelligence legislation — undefined today — this ordinance expires 91 days later. Track Indianapolis the way you track state-level AI advertising rules for preemption cliffs.

Legal uncertainty (read this before panic)

Green compliance stamp on a cream document, for uncertain enforcement of Bloomington's superintelligence ban

Analysts flagged several open questions before the vote; they remain after passage:

None of these are resolved in the November newswire. explainx.ai does not give legal advice; Bloomington companies should ask Indiana counsel whether employee ChatGPT use triggers 14.25.030 as written.

Timeline: Rollo’s third try in three years

B Square documented Rollo’s arc:

1. July 2023 — nonbinding AGI moratorium resolution withdrawn after Indiana University concerns.
2. June 2026 — unanimous existential-risk resolution (precursor to Ordinance 2026-29).
3. October 14, 2026 — ordinance introduced with five-page memo.
4. November 4, 2026 — unanimous adoption per IDS.

That pacing shows local governments can move faster than Congress on existential-risk framing, even when enforceability lags ambition.

What is confirmed versus still open

Confirmed from primary materials and IDS:

• Council passed an ordinance prohibiting superintelligence distribution and deployment in Bloomington.
• The ban is framed as interim pending state or federal rules.
• Remote access from inside the city counts as deployment.
• Civil enforcement tools exist; fiscal impact is predicted as neutral.

Open at publication:

• Mayor signature and effective date (SAFE AI Forever cited a 20-day mayoral window).
• Whether any developer receives notice or penalty.
• Court challenge timing on commerce and referendum clauses.
• How IU research and startups interpret the line between tool AI and ASI.

Related on explainx.ai

• Bernie Sanders Ban Artificial Superintelligence Act, explained
• What is superintelligence (ASI)?
• Bengio, Bannon, and the superintelligence ban coalition
• NYC Council AI validation and kill-switch hearing
• White House super-intelligence task force and Bessent
• Has AI reached superintelligence in the Astra debate?
• Anthropic pitches federal AI safety bill

Sources: Ordinance 2026-29 PDF, bloomdocs.org · Indiana Daily Student, November 6, 2026 · B Square Bulletin analysis, October 11, 2026