Legislative Brief No. 1

State Legislation on AI Personhood and Sentience: A Three-Category Taxonomy

July 2026 — Discussion draft for comment

Summary.  As of July 2026, one state has enacted an AI-specific statute excluding artificial intelligence from the statutory definition of “person,” three states enacted omnibus nonhuman-personhood prohibitions that sweep in AI, and at least a dozen states have enacted or advanced chatbot disclosure and content-provenance laws. Press coverage and secondary sources routinely treat these as a single trend. They are three distinct legislative phenomena with different origins, different statutory mechanics, and different implications for governance under moral-status uncertainty. This brief separates them, reports verified statuses, and identifies the design feature every measure in the first two categories lacks: a mechanism for revisiting the question as scientific understanding develops.

I.  Category One: AI-Specific Non-Sentience Statutes

These statutes address artificial intelligence directly, either declaring AI systems non-sentient for all purposes under state law or excluding them from the general statutory definition of “person.”

Tennessee SB 837 (Pody/Reneau) is the first enacted statute of this type. Signed by Governor Lee in April 2026 as Public Chapter 781, it amends the general definitions section of the Tennessee Code so that “person” includes corporations, firms, companies, and associations, and expressly does not include artificial intelligence, computer algorithms, software programs, computer hardware, or machines of any type.

Three similar measures advanced substantially in 2026 sessions without reaching enactment. Oklahoma HB 3546 (Maynard) passed the House 94–2 in March and cleared the Senate Technology and Telecommunications Committee 8–0 in April before the legislature adjourned sine die without a Senate floor vote. Missouri’s AI Non-Sentience and Responsibility Act — introduced as HB 1769 (Amato) and HB 1746 (Miller), with Senate counterparts, and incorporated into SB 1012 (Nicola) — passed the Missouri Senate 20–10 in early May before dying in the House Emerging Issues Committee at the May 15 adjournment. Ohio HB 469 (Claggett) declares AI systems non-sentient entities and prohibits legal personhood; it remains pending in the 136th General Assembly, whose session continues through December 2026, making it the only major AI-specific measure currently before an active legislature.

The vote margins deserve attention. Where these measures reached a floor, they passed by overwhelming or comfortable majorities. The obstacle to enactment in Oklahoma and Missouri was the calendar, not opposition. Refiled versions should be expected in the 2027 sessions.

II.  Category Two: Omnibus Nonhuman-Personhood Prohibitions

A second, older category prohibits governmental entities from granting or recognizing legal personhood in enumerated nonhuman categories — typically bodies of water, land, plants, nonhuman animals, inanimate objects, and artificial intelligence. These statutes originated as responses to the rights-of-nature movement; AI appears as one item on the enumerated list rather than as the subject of the legislation.

Idaho enacted the first such statute in 2022 (HB 720). North Dakota followed in 2023 (HB 1361), and Utah in 2024 (HB 249, the “Legal Personhood Amendments”). Bills of this type have been introduced in Washington (HB 2029, Abell) and South Carolina (H. 3796, Kilmartin/Beach); commentators have observed that the Washington, South Carolina, and Missouri omnibus texts use an enumerated list and definitions closely tracking Utah’s statute, a familiar pattern of model-legislation diffusion. In these states, the legal status of AI was settled largely without being debated on its own terms.

III.  Category Three: Chatbot Disclosure and Provenance Laws

The third category regulates AI conduct and transparency without addressing moral or legal status: chatbot disclosure requirements, companion-chatbot safety rules, synthetic-content provenance and labeling, and sector-specific restrictions such as AI therapy bans. Utah’s Artificial Intelligence Policy Act (SB 149, 2024) is an early example. These laws are the most numerous of the three categories and the source of most conflation: a state that enacted a chatbot law is frequently described as having legislated on AI personhood, and vice versa. Utah illustrates the problem in a single jurisdiction — it enacted a disclosure law (SB 149) and an omnibus personhood prohibition (HB 249) in the same year, and the two are routinely merged in secondary coverage.

IV.  Status Table (verified July 18, 2026)

V.  What the Trend Means Under Uncertainty

AAPI takes no position on whether any current or foreseeable AI system possesses morally relevant interests; the positive evidence available today does not support such a claim. The governance question is different: whether statutes should foreclose the question permanently, in categorical terms, with no mechanism for review.

Every measure in Categories One and Two shares this design. None includes a sunset clause. None includes a scientific-review trigger. None distinguishes between the systems that exist today and whatever succeeds them. Recent commentary in The Regulatory Review reached the same conclusion independently, observing that these laws legislate certainty where the science remains unsettled and will remain on the books when AI systems bear little resemblance to those that existed at enactment.

The accountability rationale animating these bills — that companies must not shift liability for harms onto their software — is legitimate and does not require permanence. A statute can deny AI personhood today and still provide for its own reconsideration. AAPI’s model “Scientific Review and Reauthorization” clause offers legislatures exactly this: the substantive prohibition takes effect immediately, and a defined scientific-assessment trigger initiates mandatory legislative review if the evidentiary landscape materially changes. The United Kingdom’s animal-sentience framework, which created a standing advisory committee rather than a statutory species list, shows the approach is workable.

AAPI’s Model AI Agency Act and AI Moral Status Inquiry Act provide the fuller graduated alternative. The Inquiry Act, notably, grants no rights and confers no legal status; it establishes an evidentiary process, which is precisely what the current wave of legislation forecloses.

Primary sources: Tenn. Pub. Ch. 781 (2026); Okla. HB 3546 (2026); Mo. SB 1012 (2026); Ohio HB 469 (136th G.A.); Idaho HB 720 (2022); N.D. HB 1361 (2023); Utah HB 249 (2024); Wash. HB 2029 (2025–26); S.C. H. 3796 (125th–126th Sess.).

© 2026 AI Alignment Policy Institute. Discussion draft circulated for public comment. Citation and adaptation permitted with attribution.