Duty LXD-4519
Deploys a mental-health chatbot, in Utah
Protection of Personal Information Act, reasonable procedures and records-destruction dutysince 4 lines
Utah Code 13-44-201 · Read the law's page
- This binds any person who conducts business in Utah and maintains personal information collected or maintained in the regular course of business.
- Implement and maintain reasonable procedures to prevent the unlawful use or disclosure of that personal information. The statute states no further content for what 'reasonable' requires beyond this general standard.
- Destroy, or arrange for the destruction of, records containing personal information that are not to be retained, by the method the section specifies.
- There is no private right of action for a violation of this section; only the Utah Attorney General may enforce this chapter.
Cybersecurity Affirmative Defense Actsince 5 lines
Utah Code 78B-4-701 to 78B-4-704 · Read the law's page
- This is a defense available to any person sued in Utah on a claim arising from a breach of system security; it creates no independent duty to adopt a cybersecurity program, and under section 78B-4-704 may not be construed to create a private cause of action, including a class action, for failing to comply with it.
- To claim the defense against a claim of failing to implement reasonable information security controls, failing to appropriately respond to a breach, or failing to appropriately notify an affected individual, have in place at the time of the breach a written cybersecurity program designed to protect the type of personal information at issue and scaled to your size, complexity, activities, and the sensitivity of the information you hold.
- Build the program as a reasonable security program, a designated coordinator, procedures to detect, prevent and respond to a breach, employee training, and periodic risk assessments of network and software design, information handling, and data storage and disposal, adjusted as circumstances change, or have it reasonably conform to a current named framework: NIST SP 800-171; NIST SP 800-53 and 800-53A; the FedRAMP Security Assessment Framework; the CIS Critical Security Controls; the ISO/IEC 27000 family; the HIPAA Security Rule or Gramm-Leach-Bliley Title V regulations for information those regimes cover; or the PCI Data Security Standard for payment card information.
- Update the program to a revised framework within one year of the revision's publication, and to an amended regulation within a reasonable time weighed against the risk to personal information and the cost and effort of compliance.
- The defense is unavailable if you had actual notice of a threat or hazard to the information's security and did not act in a reasonable amount of time to take known remedial efforts before it resulted in the breach; a risk assessment to improve security is not itself actual notice.
HB 452 (2025), mental health chatbot disclosuresince 4 lines
Utah Code 13-72a-203 · Read the law's page
- Have the chatbot clearly and conspicuously disclose that it is artificial intelligence and not human, if it simulates the kind of confidential conversation a licensed mental health therapist would have
- Disclose before the user can access the chatbot's features
- Disclose again at the start of any interaction following seven days of non-use
- Disclose again any time the user asks whether AI is being used
SB 226 (2025), required disclosures for generative AI in consumer transactionssince 4 lines
Utah Code 13-77-103 to 13-77-104 · Read the law's page
- Disclose that the individual is interacting with generative AI when they ask clearly and unambiguously, if you use generative AI to interact with someone in a consumer transaction
- Disclose proactively and prominently, verbally or in writing as applicable, if you work in a state-regulated occupation and the interaction is high risk
- High risk here means collecting sensitive health, financial, or biometric information, or giving advice relied on for significant decisions
- A safe harbor applies where the generative AI clearly and conspicuously discloses its non-human nature throughout the interaction