Utah's Artificial Intelligence Policy Act: Generative AI Disclosure, the Learning Laboratory and Mental Health Chatbots

What the 2024 Act Created

Senate Bill 149 of the 2024 General Session, titled Artificial Intelligence Amendments, was sponsored by Senator Kirk A. Cullimore and Representative Jefferson Moss. The bill status page records a 26 to 0 Senate vote on February 13, 2024, a 73 to 0 House vote on February 28, and the Governor's signature on March 13. The enrolled text took effect May 1, 2024. Later amending bills identify it as chapter 186 of the Laws of Utah 2024.

The enrolled bill did four things. It enacted a new section 13-2-12 on generative AI and consumer protection liability. It enacted an Artificial Intelligence Policy Act, which the enrolled copy placed at chapter 70 of Title 13 and which is now codified as chapter 72. It amended the Title 13 repeal list so that the new chapter would expire on May 1, 2025. And it enacted section 76-2-107 of the criminal code, under which an actor may be found guilty of an offense committed with the aid of generative AI, or one the actor intentionally prompts or otherwise causes generative AI to commit.

The 2024 version of section 13-2-12 defined generative AI as an artificial system trained on data that interacts with a person through text, audio or visual communication and generates non-scripted, human-like output with limited or no human oversight. It then imposed two disclosure duties. Any person who used generative AI to interact with someone in connection with an act the Division of Consumer Protection administers had to disclose clearly and conspicuously, if asked or prompted, that the person was dealing with generative AI and not a human. And a person providing the services of a regulated occupation had to disclose prominently, in every case, when a client was interacting with generative AI, verbally at the start of an oral exchange and through electronic messaging before a written one.

Generative AI Disclosure on Request and in Regulated Occupations

The disclosure rules now sit in chapter 77 of Title 13, Generative Artificial Intelligence: Consumer Disclosures and Enforcement. Section 13-77-103 sets out two duties that apply to different people:

• A supplier that uses generative AI to interact with an individual in connection with a consumer transaction discloses that the individual is interacting with generative AI and not a human, if the individual asks or otherwise prompts the supplier about whether AI is being used
• That prompt has to be "a clear and unambiguous request to determine whether the interaction is with a human or with artificial intelligence"
• An individual providing services in a regulated occupation prominently discloses the use of generative AI when that use constitutes a high-risk artificial intelligence interaction, and remains subject to every requirement of the occupation when providing services through generative AI
• The professional's disclosure is given verbally at the start of a verbal interaction and in writing before the start of a written one

The definitions in section 13-77-101 carry most of the weight. Generative AI is now an AI technology system trained on data, designed to simulate human conversation with a consumer through text, audio or visual communication, and generating non-scripted output. A regulated occupation is one regulated by the Department of Commerce that requires a license or state certification. A high-risk artificial intelligence interaction is one that involves collecting sensitive personal information such as health, financial or biometric data; providing personalized financial, legal, medical or mental health advice or services that could reasonably be relied on for significant personal decisions; or another application the Division defines by rule.

Section 13-77-104 adds a safe harbor. A person is not subject to an enforcement action under section 13-77-103 if its generative AI clearly and conspicuously discloses, at the outset of any interaction in a consumer transaction or regulated service and throughout it, that it is generative AI, is not human, or is an AI assistant. The Division, in consultation with the Office of Artificial Intelligence Policy, may make rules identifying forms and methods of disclosure that do or do not satisfy that standard.

The 2025 Amendments and the Narrowed Disclosure Duty

Three bills in the 2025 General Session rewrote the 2024 structure. SB 226, Artificial Intelligence Consumer Protection Amendments, again sponsored by Senator Cullimore and Representative Moss, passed the Senate 24 to 0 and the House 69 to 0 and was signed on March 27, 2025. The enrolled bill repealed section 13-2-12 and enacted the six sections now codified as chapter 77, effective May 7, 2025; the codified notes record it as chapter 465 of the 2025 session. The enrolled copy numbered the new chapter 75, and the codifiers renumbered it to avoid a collision with other 2025 enactments.

Read side by side, the two versions differ on scope and trigger. The 2024 request-based duty reached anyone using generative AI in connection with any act the Division administers; the 2025 duty reaches a supplier in a consumer transaction, and the request must be clear and unambiguous. The 2024 professional duty applied to every interaction with generative AI in regulated services; the 2025 duty applies only to high-risk interactions. The 2025 bill also added the safe harbor and made a violation of the chapter a violation of the Consumer Sales Practices Act's deceptive-act provision.

SB 226 and a separate bill, SB 332, Artificial Intelligence Revisions, each moved the chapter's repeal date from May 1, 2025 to July 1, 2027. SB 332 passed the Senate 24 to 0 and the House 74 to 0 and was signed on March 25, 2025. The third bill, HB 452, is described below. In the 2026 General Session, HB 320, Office of Artificial Intelligence Policy Amendments, sponsored by Representative Paul A. Cutler and Senator Cullimore and signed March 18, 2026, rewrote the office's program; its provisions are in force from May 6, 2026.

The Office of Artificial Intelligence Policy and Regulatory Mitigation

Section 13-72-201 creates the Office of Artificial Intelligence Policy in the Department of Commerce, with a director appointed by the department's executive director. The office administers a learning laboratory program, consults with businesses about potential regulatory proposals, makes rules on participant disclosures to consumers and reporting, and reports each year before November 30 to the Business and Labor Interim Committee on its learning agenda, outcomes, agreements and recommended legislation. Under section 13-72-301, the laboratory exists to analyze the risks, benefits and policy implications of AI technologies, to encourage responsible deployment, and to produce findings and recommendations.

The relief mechanism is in part 4, which HB 320 renumbered from sections 13-72-302, 303 and 305, while repealing former section 13-72-304. Section 13-72-401 lets a participant apply for a regulatory mitigation agreement or, a category HB 320 added, a joint interpretation agreement that clarifies how an existing law or rule applies to an AI technology. A regulatory mitigation agreement, as section 13-72-101 defines it, permits use of an AI technology despite a law or rule that might impede it and may set cure periods, reduced civil fines or tailored safeguards. Each agreement specifies limits on users and geography, safeguards, consumer disclosures and audit reporting, and the office audits participants while it is in effect.

Eligibility under section 13-72-402 turns on technical expertise, financial resources, potential consumer benefit that may outweigh the risks of mitigated enforcement, a risk monitoring plan and a limited test scale. Under section 13-72-403, an initial demonstration period may not exceed 12 months, and the office may grant up to two 12-month extensions. The office may terminate an agreement at any time for any reason, the participant stays bound by every requirement the agreement does not expressly waive, and participation is not a state endorsement.

The office's register of authorized pilots, retrieved for this article, lists agreements co-signed with the Division of Professional Licensing. The earliest is ElizaChat, signed November 18, 2024 for a year, covering a student emotional support app; the register describes it as granting a 30-day period to remedy any conversation that crosses into licensed mental health therapy. Later entries include Dentacor (May 31, 2025, dental hygiene with an AI radiograph tool), Doctronic (October 24, 2025, prescription renewals), Nolla Health (September 22, 2026, topical acne prescriptions), Expect Fitness and August AI (both October 2, 2026), and a master agreement with University of Utah Health signed September 10, 2026 under which no pilot had yet been approved.

Mental Health Chatbots

HB 452, Artificial Intelligence Amendments, sponsored by Representative Moss and Senator Cullimore, passed the House 70 to 0 and the Senate 28 to 0, was signed March 25, 2025, and took effect May 7, 2025. It enacted chapter 72a. A mental health chatbot is AI technology that uses generative AI to hold conversations similar to the confidential communications an individual would have with a licensed mental health therapist, and that a supplier represents, or a reasonable person would believe, can provide therapy or help manage or treat mental health conditions. Tools that only provide scripted output, such as guided meditations, or that only route a person to a human therapist, are excluded.

• Section 13-72a-201: a supplier may not sell to or share with any third party a Utah user's individually identifiable health information or user input, except health information requested by a health care provider with the user's consent, provided to the user's health plan at the user's request, or shared with a contractor for functionality under HIPAA's privacy and security rules as if the supplier were a covered entity
• Section 13-72a-202: advertising inside a chatbot conversation must be labeled and must disclose any sponsorship or affiliation, and a user's input may not be used to decide whether to show an ad, what to advertise, or how to present it
• Section 13-72a-203: the chatbot discloses that it is AI and not a human before the user may access its features, at the start of any interaction if the user has not accessed it in the previous seven days, and whenever the user asks

HB 452 also enacted section 58-60-118 in the mental health professional practice act. It creates an affirmative defense, in an action under section 58-1-501(1) or (2), for a supplier that maintains documentation of its foundation models, training data and data practices, files a written policy with the Division, and complies with it. The policy must describe, among fifteen listed procedures, how licensed therapists take part in development, how the chatbot is tested before release against the risk posed in therapy with a licensed therapist, how it responds in real time to acute risk of physical harm, and how it prioritizes user mental health and safety over engagement metrics or profit.

Enforcement by the Division of Consumer Protection

Section 13-77-102 carries forward the 2024 liability rule: it is not a defense to a violation of any statute the Division administers that generative AI made the violative statement, undertook the violative act, or was used in furtherance of the violation. Under section 13-77-105, a violation of chapter 77 is a violation of section 13-11-4(1), the deceptive-act provision of the Consumer Sales Practices Act. The Division director may impose an administrative fine of up to $2,500 per violation, and in a Division action a court may declare the practice unlawful, enjoin it, order disgorgement and repayment, and impose a fine of up to $2,500 per violation. Violating an administrative or court order carries a civil penalty of up to $5,000 per violation, and the Attorney General acts as the Division's counsel.

Section 13-72a-204 gives the Division the same fine and court remedies for the chatbot chapter, with collected money going to the Consumer Protection Education and Training Fund. Section 13-77-106 states that chapter 77 displaces no other state or federal remedy. This article reports no enforcement action under chapter 77, chapter 72a or former section 13-2-12; none appeared in the primary sources reviewed.

The sunset falls unevenly. Section 63I-2-213 provides that chapter 72, the Artificial Intelligence Policy Act, is repealed July 1, 2027, which would end the office and the learning laboratory unless the Legislature acts. The section names no repeal date for chapter 77. The le.utah.gov pages for chapter 72a carry a note that they are affected by section 63I-2-213 on July 1, 2027, although the text of that section names only chapter 72.

Frequently Asked Questions