In force since May 1, 2024 · State governance
SB 149 · Artificial Intelligence Policy Act
The 2024 act created the Office of Artificial Intelligence Policy and an Artificial Intelligence Learning Laboratory. The framework gives the office a structured route to study AI risks and benefits and, under statutory conditions, enter agreements that clarify legal interpretations or mitigate specified regulatory requirements during a limited demonstration period. SB 332 and SB 226 extended the act's repeal date to July 1, 2027. HB 320 revised the operating model in 2026.
Who should read it: AI developers and deployers seeking a Utah agreement, state agencies, governmental entities, and policy teams. Important limit: participation is not state endorsement, and obligations outside an agreement remain applicable.
2024 enrolled act ↗ · current Utah Code chapter ↗
In force since May 7, 2025 · Consumer protection and professional services
SB 226 · generative AI disclosures and enforcement
The act makes clear that generative AI is not a defense to conduct that violates a consumer protection law. A supplier using generative AI in a consumer transaction must disclose that the interaction is not human when an individual makes a clear and unambiguous inquiry. A person providing regulated occupational services must give a prominent disclosure before a high risk AI interaction. The statute also provides an outset and continuing disclosure safe harbor, while preserving the underlying requirements of the regulated occupation.
Exposure: Division of Consumer Protection enforcement, administrative or judicial fines of up to $2,500 per violation, and up to $5,000 for violating an administrative or court order, subject to the statute's terms.
Enrolled act and effective clause ↗ · current Utah Code chapter ↗
In force since May 7, 2025 · Mental health chatbots
HB 452 · mental health chatbot protections
The act regulates a defined class of chatbots represented, or reasonably understood, as providing mental health therapy or help managing or treating a mental health condition. It restricts the sale and sharing of user inputs and identifiable health information, limits targeted advertising based on user inputs, and requires recurring disclosure that the chatbot is AI and not human. The Division of Consumer Protection enforces the chapter.
A separate provision creates a limited affirmative defense to specified unlicensed practice claims when a supplier files and follows a documented policy covering clinical participation, testing, foreseeable harms, acute risk protocols, review, discrimination, privacy, and user safety. Filing a policy is not licensure and does not prevent an enforcement action.
Enrolled act and effective clause ↗ · current Utah Code chapter ↗
In force since May 6, 2026 · State governance
HB 320 · policy office and learning laboratory revisions
The 2026 amendment reframed the learning laboratory around responsible deployment and added joint interpretation agreements alongside regulatory mitigation agreements. It extended the agreement process to relevant agencies and governmental entities, required written safeguards, consumer disclosures, reporting, and regular audits, and allowed up to two 12 month extensions after an initial period of no more than 12 months.
Analytic significance: Utah is using supervised, time limited agreements as an institutional learning mechanism. The statute does not create a blanket AI exemption.
Enrolled act and May 6 effective date ↗
In force since July 1, 2026 · Public education
HB 273 · classroom AI use policies
The act requires the State Board of Education to publish a model AI use policy and requires local education agencies to adopt policies based on it. The statutory model retains educator judgment, prohibits generative AI from independently grading student work or making high stakes student determinations, requires approval for student facing tools, calls for parent notice when generative AI is used in classroom instruction or assessment, and bars specified uses such as biometric surveillance and psychological profiling.
The act also establishes standards for optional high school AI sandbox courses and requires safeguards, parent notice, and written consent. Some implementation milestones occur after the act's July 1, 2026 effective date, including a December 1, 2026 model policy deadline and a July 1, 2027 local policy deadline.
Enrolled act, effective clause, and implementation dates ↗ · Official status and history ↗