S2195 – Artificial Intelligence Companion Models

RILawState

Date Passed

6/22/2026

Effective Date

1/1/2027

Summary

Regulates operators of AI companion systems that use AI, generative AI, or emotional-recognition algorithms to simulate sustained human-like relationships with users. Requires covered AI companions to include protocols for suicidal ideation, self-harm, and possible physical harm to others, provide crisis-resource notifications when those expressions are detected, provide recurring disclosures that the user is not communicating with a human, and file annual safety-protocol reports with the Attorney General beginning July 1, 2027. Authorizes AG enforcement and civil penalties.

Healthcare Implications

Not healthcare-specific, but relevant to mental-health-adjacent AI companions and consumer AI systems that may interact with users expressing suicidality, self-harm, or emotional distress. Operators must implement crisis-response protocols, provide human/non-human disclosure, report safety-protocol metrics to the Attorney General, and face civil penalties for noncompliance.

Operational Implications

  • Operators must provide a clear and conspicuous notification at the beginning of any AI companion interaction and at least every three hours during continuing interactions that the user is not communicating with a human.
  • Operators may not provide an AI companion unless it contains a protocol for possible suicidal ideation, self-harm, or physical harm to others and provides crisis-resource notifications as soon as those expressions are detected.
  • Beginning July 1, 2027, operators must file annual reports with the Attorney General that include the number of safety-protocol activations and related metrics; the Attorney General must publish aggregated data.

Impact Level

Medium

Keywords

Safety & Risk; Transparency & Governance; Clinical Quality & Efficacy

Stakeholders

Patients & Public; Developers & Vendors; Regulators & Government