Rhode Island limits how therapy providers use artificial intelligence

Therapy offered to the public must be conducted by a licensed provider. Providers cannot let AI make independent treatment decisions, treatment recommendations, or treatment plans. The rules cover therapy offered to Rhode Island's public, including online AI services. AI tools cleared for healthcare use by the Food and Drug Administration or another responsible federal agency are exempt, as is qualifying research. Direct AI therapy communication requires an established provider relationship and patient consent. An in-person visit is not required. Providers retain responsibility for care decisions and reasonable oversight of patient AI use, but not vendor-controlled designs, algorithms, or outputs. The chapter excludes faith-based religious counseling, peer support, and public self-help or educational resources that do not claim to offer therapy. Research must have a healthcare facility review board's approval and meet the cited federal requirements for patient consent, removal of identifying details, or both to qualify for exemption. The law takes effect upon passage.

Providers must explain covered AI use in writing and obtain specific written consent. This requirement covers AI companions for mental or emotional support and AI designed to build emotional attachment, bonding, or dependency, used for therapy support or treatment communication in recorded or transcribed sessions. The patient or an authorized representative must receive written notice that AI will be used and its specific purpose. Consent must be informed, voluntary, specific, and revocable. Electronic written consent counts; general terms of use or deceptive actions do not. The chapter exempts federally cleared healthcare AI tools and research meeting its review-board approval and federal consent or identity-removal requirements.

Providers must protect therapy information. Rhode Island's Executive Office of Health and Human Services may investigate violations and issue rules to carry out the law. Privacy duties cover provider-held records and patient-provider communications. Disclosure remains allowed under the referenced state-law exceptions. When clients disclose using AI on their own, providers may offer guidance and must discuss risks, monitor safety, and intervene when necessary. Privacy violations carry penalties under the referenced state confidentiality law. The chapter's exemptions apply, including federally cleared healthcare AI tools and research meeting its approval and federal consent or identity-removal requirements.