Senate Bill S.788 would require patient consent before AI assists in therapy; committee adopts amendments and reports it favorably

Get AI-powered insights, summaries, and transcripts

Sign Up Free
AI-Generated Content: All content on this page was generated by AI to highlight key points from the meeting. For complete details and context, we recommend watching the full video. so we can fix them.

Summary

The Labor, Commerce and Industry Committee advanced S.788, which would require licensed health professionals to get written informed consent before using AI in therapy or psychotherapy and forbids AI from making independent therapeutic decisions; the committee adopted a subcommittee amendment to create a new Chapter 86 and approved the bill as amended after debate over definition scope, real-time transcription and open versus closed AI.

The Labor, Commerce and Industry Committee advanced Senate Bill S.788 after adopting a subcommittee amendment and a later perfecting amendment, sending the bill out of committee with a favorable report.

Sen. Blackman (Lancaster), the bill’s sponsor, said he introduced S.788 after reviewing the rapid expansion of artificial intelligence and the need for “guardrails” in psychotherapy and psychology. Committee staff member Rebecca summarized the measure’s main provisions: it would add an article to Chapter 1 of Title 40 with definitions and prohibit a health professional from using artificial intelligence in therapy or psychotherapy without the patient’s informed consent; the patient must be told the AI’s use and purpose, AI may not make independent therapeutic decisions, records between a licensed professional and client must remain confidential, and the appropriate licensing board could investigate violations and impose fines up to $10,000.

The committee adopted a subcommittee amendment recommended by the legislative drafting office (LLR) to create a new Chapter 86 specifically addressing AI use in therapy and to add corresponding language into the practice acts for listed licensed professionals so the protections appear directly in each practice act.

Members debated the bill’s scope. Senator from Edgefield said the intent is to ensure AI is not used in the provider–patient interaction without the patient’s knowledge. Senator from Dorchester asked whether the bill distinguishes recording from transcription (real-time transcription vs. creating a retained recording) and whether the requirement for disclosure and consent was intended to be written consent. Rebecca said the text, as drafted, would require the information be provided in writing and would require written consent. Committee discussion also focused on open-source versus closed-source AI tools. Michael Johnson noted professional-ethics guidance from bar associations regarding open systems and cautioned about sharing client-identifying information with open models; other senators said the bill was meant to cover closed-setting uses tied to clinical records and direct care.

The chair raised a separate concern about the bill’s definition of therapy potentially being overbroad (for example, capturing individual internet users who ask general medical questions of public chatbots). He proposed adding wording limiting coverage to interactions with the intent to diagnose; the chair observed that the measure should target the provider–patient relationship rather than incidental public use of chatbots. The committee considered and approved a clarifying amendment along those lines (a perfecting amendment), and then voted to report the bill favorably as amended.

Throughout the hearing senators emphasized that licensing boards would retain investigatory authority and that continuing work on AI policy would likely be necessary. The committee completed its action by reporting S.788 out of committee with the adopted subcommittee and perfecting amendments; recorded tallies were taken by voice vote in the transcript.