Statehouses Close Their Sessions With a Dense Wave of AI Laws
· TexaHub · Regulation & Ethics, Artificial Intelligence, Geopolitics & Tech
Between late May and early June, Connecticut, Illinois, Colorado and New York advanced frontier-model, chatbot, transparency and data center bills despite federal pressure against state AI rules.
Two weeks, four states
The end of the spring legislative calendar has produced one of the busiest stretches of state AI lawmaking so far. Connecticut Governor Ned Lamont signed the state's omnibus AI and online safety bill, SB 5, on 27 May [1]. Illinois lawmakers gave final approval to SB 315, a frontier-model safety bill, and sent it to Governor JB Pritzker on 1 June together with four other AI-related bills [2]. In Colorado, Governor Polis signed a chatbot safety bill, HB 1263, on 29 May and a bill on AI in health care approvals, HB 1139, on 2 June [2]. New York's legislature then closed its session by passing a package of AI bills and a one-year moratorium on large data center permits [3].
Smaller measures moved in parallel. Louisiana Governor Landry signed HB 369, on AI disclosure in campaign phone communications, on 29 May, and the New York Senate and Assembly approved S 9408A, a ban on selling AI chatbot toys, on 1 and 2 June [2]. Colorado's 2024 AI Act was itself revised by SB 189, signed on 14 May [2].
The common thread is that states are filling a federal gap on their own terms, and they are doing so in at least four distinct regulatory lanes: frontier-model safety, chatbots and minors, transparency about training data and synthetic content, and the physical infrastructure that AI depends on. Each lane carries different compliance consequences for developers and operators.
Illinois joins the frontier-model club
SB 315 passed the Illinois Senate 52-5 and the House 110-0 [4] [5]. It applies to large developers with more than $500 million in revenue and requires them to adopt transparency frameworks, report on the catastrophic-risk capabilities of their models and submit to third-party audits of compliance [4]. Its effective date was pushed back to 2028 during negotiations [4] [5]. The Illinois attorney general would have exclusive enforcement authority, with civil penalties of up to $3 million per violation and no private right of action [5].
Capitol News Illinois reported that OpenAI and Anthropic backed the bill, with Cesar Fernandez of Anthropic saying that "as these models grow more powerful, this kind of enforceable accountability matters more than ever" [5]. The TechNet industry coalition objected to what it described as subjective determinations in the absence of national standards [5]. Earlier in the debate, Jeremy Kudon of the American Innovators Network argued that the audits were an expensive requirement for startups, which would have to spend capital on lawyers and lobbyists simply to comply [4]. Pritzker has pledged to sign the bill [1].
If signed, Illinois would become the third state to regulate frontier models, after California and New York in 2025 [5]. New York's RAISE Act, signed in December, applies to developers with more than $500 million in revenue, requires safety incidents to be reported within 72 hours and takes effect on 1 January 2027, with penalties of up to $1 million for a first violation and $3 million for later ones [6]. The repeated $500 million revenue threshold suggests that a de facto multistate template for frontier-model oversight is forming, with Illinois distinguishing itself mainly through mandatory third-party audits.
Connecticut's omnibus and the chatbot lane
Connecticut chose breadth over a single-issue statute. According to the Transparency Coalition, the signed law includes consumer disclosures from operators of subscription-based AI products and safety obligations and whistleblower protections for employees of frontier AI developers [1]. The group's pre-signing guide to the bill described chatbot rules taking effect on 1 January 2027, including hourly disclosure that the user is talking to AI and safety protocols for minors, along with provenance data requirements for developers with more than 1 million monthly users [7]. The same guide said frontier developers with more than $500 million in annual revenue must set up anonymous channels for reporting catastrophic risks by 1 January 2027, and that AI hiring disclosures would apply from 1 October 2027 [7].
Chatbots aimed at or used by minors are the most crowded lane. New York's S 9051, sponsored by Senator Kristen Gonzalez and Assemblymember Alex Bores, passed 137-0 in the Assembly and 60-0 in the Senate; it bars companion chatbot operators from offering a list of unsafe features to users under 18, such as suggesting the chatbot experiences human emotions or facilitating self-harm, with penalties of up to $25,000 per violation enforced by the attorney general [3]. Colorado's HB 1263 and its therapy chatbot measure, HB 1195, point the same way [2].
In our view, the near-unanimous votes on child-focused chatbot bills contrast with the narrower margins on frontier-model rules, which suggests that minor protection is the area where state rules are least likely to be contested politically.
New York's transparency package
New York also passed the AI Training Data Transparency Act, A 6578, which cleared the Senate 54-6 on 4 June and would require generative AI developers to "post a high-level summary of the datasets used in the development of the AI model or service" [3]. A provenance bill, S 6954, passed 60-1 in the Senate and 141-0 in the Assembly and would require synthetic content to carry provenance data identifying that it was AI-made and by which provider, taking effect 180 days after enactment [3].
The FAIR News Act, S 8451, would require conspicuous disclosure on news content substantially created by AI, with penalties of $1,000 for a first offense and $5,000 for later ones [3]. The Assembly also passed a surveillance pricing ban, A 9349, by 100-42 [3]. Governor Hochul has until 31 December 2026 to act on the bills [3].
Hochul's handling of the RAISE Act is the closest precedent. The version she signed reduced developer obligations compared with the bill the legislature passed, including cutting the maximum first-offense penalty from $10 million to $1 million [6]. A long signing window and the governor's willingness to negotiate amendments mean the final shape of the training-data and provenance rules remains open.
A first data center moratorium
The most consequential bill for infrastructure is the Responsible Data Center Development Act, sponsored in the Senate by Gonzalez, which passed the Senate 43-17 on 4 June [8] and the Assembly 103-38 [3]. The bill "places a one year moratorium on the issuance of data center permits" and then requires public hearings in host communities [8]. It would also require utilities to create separate rate classes for data centers of 20 megawatts or more so that these facilities cover the cost of grid upgrades, and it sets renewable energy targets of 33% by 2034, 67% by 2039 and 90% by 2040 [8].
Environmental groups framed the vote around household energy costs. Liz Moran of Earthjustice said that "1 in 4 New Yorkers already can't afford their energy bills, and that's without the rapid buildout of AI data centers" [9]. Food & Water Watch said New York would be the first state to enact such a temporary halt if Hochul signs, and noted that a similar bill passed Maine's legislature but was vetoed by Governor Janet Mills [10].
For operators planning capacity in the US Northeast, the moratorium is the item to watch most closely. Even without a signature, the bill's cost-allocation and renewable sourcing provisions indicate the conditions that state legislators may attach to large loads in other states, and they shift part of the AI infrastructure debate from federal energy policy to state permitting.
Federal pressure and what comes next
All of this is happening under an executive order signed on 11 December 2025 that calls for a "minimally burdensome national standard" for AI [11]. The order directed the attorney general to set up an AI Litigation Task Force within 30 days to challenge state AI laws, gave the Commerce Secretary 90 days to identify problematic state laws, and raised the prospect that states with such laws could lose remaining BEAD broadband funds [11]. It also listed areas that proposed federal legislation should not preempt, including "child safety protections" and "AI compute and data center infrastructure, other than generally applicable permitting reforms" [11]. Illinois legislators moved ahead despite the order [4].
Those carve-outs matter for reading this session. Chatbot rules for minors and New York's data center moratorium sit in categories that the order itself treats as outside the scope of federal preemption, while frontier-model safety and training-data disclosure rules do not. Our assessment is that the latter group is where any federal legal challenge is most likely to focus, although we are not aware of any challenge filed against the bills described here, and most of them are not yet law.
The near-term calendar is set by governors rather than legislatures: Pritzker's decision on SB 315, Hochul's on the New York package before the end of the year, and the remaining sessions in states such as California, where 30 AI bills are still in second-chamber committees ahead of a 2 July summer adjournment deadline [2]. For compliance teams, the practical takeaway is that the volume of state obligations is rising faster than any federal standard is taking shape, and that most of the new rules cluster around a small number of shared definitions, such as revenue thresholds, minors and synthetic content, that can be tracked across jurisdictions.
Sources
[1] AI Legislative Update: May 29, 2026 — Transparency Coalition, 28 May 2026
[2] AI Legislative Update: June 5, 2026 — Transparency Coalition, 4 Jun 2026
[3] New York lawmakers wrap up by passing kids chatbot safety bill and two AI transparency acts — Transparency Coalition, 9 Jun 2026
[4] Bill regulating powerful AI models advances as advocates say it's only the first step — Capitol News Illinois, 22 May 2026
[5] Illinois lawmakers pass landmark AI accountability bill — Capitol News Illinois, 27 May 2026
[6] Hochul enacts New York's AI safety and transparency bill — IAPP, 22 Dec 2025
[7] TCAI Bill Guide: SB 5, Connecticut's omnibus AI and online safety bill — Transparency Coalition, 12 May 2026
[8] Senate Bill S10642: Responsible Data Center Development Act — New York State Senate, undated
[9] NYS Legislature Passes Data Center Moratorium — Earthjustice, 5 Jun 2026
[10] NY Gov. Hochul Must Sign One-Year AI Data Center Moratorium Passed by Legislature — Food & Water Watch, 4 Jun 2026
[11] Ensuring a National Policy Framework for Artificial Intelligence — The White House, 11 Dec 2025