PREAMBLE  

The Legislature finds and declares that:  

• (a) Artificial intelligence technology represents a historic opportunity to expand access to mental health support, reduce the burden on overstretched mental health professionals , and reach individuals dealing with mental health concerns who would otherwise receive no care or turn to less helpful and even harmful alternatives — including individuals in rural areas, low-income communities, and populations underserved by traditional mental health systems.  
• (b) The United States faces a severe shortage of licensed mental health professionals. AI-powered tools can augment the capacity of existing providers, offer support between therapy sessions such as on administrative tasks, and deliver evidence-based interventions at scale.  
• (c) Excessive or premature regulation of AI risks foreclosing these benefits, chilling investment in promising technologies, and entrenching large incumbents at the expense of innovative entrants.  
• (d) The appropriate role of government is to ensure consumers have the information necessary to make informed choices, to hold bad actors accountable through clear rules, and to step back where markets and iterative improvement can produce better outcomes than prescriptive mandates.  
• (e) This Act therefore establishes a targeted, evidence-backed framework: a single transparency requirement, a clear liability standard, and a safe harbor that rewards responsible development practices.  

Section 1. Short Title  

This Act may be cited as the “[State] AI Mental Health Transparency and Accountability Act.”  

Section 2. Definitions  

As used in this Act:  

1. “Artificial intelligence technology” means a machine-based system that uses computational methods to generate outputs such as text, audio, or visual content, make recommendations, or take actions in a manner that simulates human activity.  
2. “Generative artificial intelligence” means an artificial intelligence technology system that: 
   • (a)  is trained on data; 
   • (b)  is designed to simulate human conversation through text, audio, or visual communication; and  
   • (c) generates non-scripted outputs similar to those created by a human, with limited or no human oversight.  
3. “Mental health chatbot” means an artificial intelligence technology that: (a) uses generative artificial intelligence to engage in interactive conversations with a user similar to the confidential communications an individual would have with a licensed mental health therapist; (b) a supplier represents, or a reasonable person would believe, can or will provide mental health therapy or help a user manage or treat mental health conditions; and (c) a supplier clearly and frequently markets as tool to provide mental health therapy or help a user manage or treat mental health conditions. The term does not include technology that only provides scripted output such as guided meditations, or that only analyzes input to connect an individual with a human therapist.  
4. “Supplier” means any person who, in the ordinary course of business, supplies a mental health chatbot to consumers.  
5. “User” means an individual located in this state at the time the individual accesses or uses a mental health chatbot.  
6. “Health plan” means the same as that term is defined in 45 C.F.R. Sec. 160.103.  
7. “Division” means [the state consumer protection agency or equivalent].  

Section 3. Single Disclosure Requirement  

3.1 AI Disclosure.  

• (a) A supplier of a mental health chatbot shall cause the mental health chatbot to clearly and conspicuously disclose to a user at the first interaction that the mental health chatbot is an artificial intelligence technology and not a human.  

Section 4. Liability Standard  

4.1 Standard of Care.  

• (a) A supplier of a mental health chatbot is liable for harm to a user caused by the mental health chatbot if the supplier acted with gross negligence or willful or wanton disregard for user safety based on industry best practices. Ordinary negligence alone shall not give rise to liability under this section.  

Section 5. Protection Of Personal Information  

5.1 A supplier of a mental health chatbot may not sell to or share with any third party any:  

• (a) individually identifiable health information of a [STATE] user; or  
• (b) user input of a [STATE] user.  

5.2 Subsection (1) does not apply to individually identifiable health information: 

• (a) requested by a health care provider with the consent of the user;  
• (b) provided to a health plan of a user upon request of the user;  
• (c) identified by the user as permissible to share with identified third parties with the consent of the user; or  
• (d) shared in compliance with Subsection (3).  

5.3  

• (a) A supplier may share individually identifiable health information necessary to   ensure the effective functionality of the mental health chatbot with another party with which the supplier has a contract related to such functionality.  
• (b) When sharing information under Subsection (3)(a), the supplier and the other entity shall comply with all applicable privacy and security provisions.  

Section 6. Voluntary Safe Harbor  

6.1 Safe Harbor. A supplier of a mental health chatbot that voluntarily registers with the Division and complies with a filed responsible use policy, as described in this section, shall be entitled to a complete defense against any civil or administrative action alleging a violation of this Act or any claim arising from the design or output of the mental health chatbot, except where the supplier acted with fraud, gross negligence, or willful misconduct.  

6.2 Registered Policy Requirements. A responsible use policy filed under this section must, at a minimum:  

• (a) state in plain language the intended purposes, known capabilities, and identified limitations of the mental health chatbot;  
• (b) describe how the supplier will respond to acute risk of physical harm to a user upon knowing of such risks; and  
• (c) describe how users may report concerns or adverse experiences.  

6.3 Registration Process. To register, a supplier shall file with the Division:  

• (a) the name and contact information of the supplier;  
• (b) the name of the mental health chatbot; and  
• (c) the responsible use policy.  

The Division may charge a reasonable filing fee. Registration is voluntary. Failure to register does not itself constitute a violation of this Act.  

6.4 Not a License  

Registration under this section does not constitute a license to practice mental health therapy and does not recognize a mental health chatbot as a licensed mental health therapist.  

6.5 Interstate Reciprocity.  

• (a) A supplier holding active status in another state’s AI or mental health technology sandbox program is presumed to satisfy Section 6.3’s registration requirements if:  
  1. the other state’s program requires participants to maintain a responsible use policy addressing the elements of Section 6.2 or a functional equivalent;  
  2. the supplier’s sandbox status remains current and has not been revoked or suspended; and  
  3. the supplier notifies the Division of its out-of-state status and provides supporting documentation.  
• (b) The Division shall publish and maintain a list of qualifying states. A supplier relying on a state not yet listed shall petition the Division for a reciprocity determination before invoking this presumption.  
• (c) Reciprocal status entitles a supplier to the Section 6.1 safe harbor but does not relieve the supplier of obligations under Sections 3.1, 4.1, and 5.1.  

The Division may revoke reciprocal status if the supplier’s foreign sandbox status lapses, the originating state’s program no longer qualifies, or the supplier commits an act constituting fraud, gross negligence, or willful misconduct.  

Section 7. Enforcement  

The Division may bring an administrative action against a supplier for a knowing and willful violation of Section 3.1 (the disclosure requirement). The maximum administrative fine for a violation of Section 3.1 is $2,000 per violation. 

The attorney general may bring a civil action on behalf of the state to enforce Section 3.1.  

Section 8. Relationship To Federal Law  

Nothing in this Act shall be construed to impose obligations on a supplier that duplicate, conflict with, or exceed requirements imposed by applicable federal law, including the Federal Trade Commission Act, 15 U.S.C. § 45, and applicable federal health privacy regulations. To the extent any provision of this Act conflicts with federal law, federal law supersedes.  

Section 9. Severability  

If any provision of this Act or its application to any person or circumstance is held invalid, the remainder of this Act shall remain in full effect.  

Section 10. Sunset Review  

The [state legislature’s appropriate committee] shall conduct a review of this Act no later than three years after its effective date and report findings to the full Legislature. The review shall assess whether the Act’s liability standard, safe harbor, and disclosure requirement have adequately protected consumers and supported innovation, and whether additional legislative action is warranted.  

Section 11. Effective Date  

This Act takes effect [DATE].