Rhode Island Now Requires AI Companion Bots to Run Crisis Protocols and Report to the Attorney General
Regulatory summary: Rhode Island has enacted S2195 Substitute A, a new chapter of commercial law that forces AI companion operators to build self-harm and violence-detection protocols, disclose that users are not talking to a human, and file annual activation reports with the attorney general. It takes effect January 1, 2027.
By Anthony Guerriero, Founder, The Leveraged Years · Reviewed by The Leveraged Years Editorial Desk · Published July 9, 2026 · Last updated July 9, 2026
The Leveraged Years AI Regulation Tracker
Key takeaways
Rhode Island moved from having no companion-bot safety rule to a binding duty. Operators must now embed a crisis protocol, deliver a repeated non-human disclosure, and prepare to report activations to the attorney general. Before this act, none of these were legal requirements in the state.
Operators and product teams behind AI companion and character-style chat apps; compliance and legal officers at those companies; executives who own product risk; and mental-health and wellness technology vendors whose tools sustain personal, emotionally responsive dialogue with consumers.
Status: Signed into law June 22, 2026.
Determine whether your product meets the AI companion definition, then scope a crisis-protocol and disclosure build with a hard January 1, 2027 deadline and an activation-metrics logging layer ready for the first attorney general report in mid 2027.
Date
Jurisdiction
Rule
Affected professionals
Status or effective date
2026-07-09
United States (Rhode Island)
Rhode Island moved from having no companion-bot safety rule to a binding duty. Operators must now embed a crisis protocol, deliver a repeated non-human disclosure, and prepare to report activations to the attorney general. Before this act, none of these were legal requirements in the state.
Operators and product teams behind AI companion and character-style chat apps; compliance and legal officers at those companies; executives who own product risk; and mental-health and wellness technology vendors whose tools sustain personal, emotionally responsive dialogue with consumers.
Signed into law June 22, 2026. Effective January 1, 2027. Attorney general reporting obligation begins July 1, 2027.
Frequently Asked Questions
Is Rhode Island S2195 in force now?
The act was signed on June 22, 2026 and is enacted statute, but its operating and disclosure duties take effect January 1, 2027. The obligation to file annual reports with the attorney general begins July 1, 2027.
Does the law cover all chatbots?
No. It applies to AI companions, defined as systems that simulate a sustained human-like relationship by retaining prior interactions, asking unprompted emotion-based questions, and sustaining personal dialogue. It expressly excludes customer service bots, tools marketed for efficiency, research, or technical assistance, and internal or employee productivity systems.
What must the crisis protocol do?
It must address possible suicidal ideation or self-harm and possible physical harm to others expressed by a user, and it must notify the user of crisis services such as a suicide hotline or crisis text line as soon as those expressions are detected.
How often must the non-human disclosure appear?
At the beginning of any AI companion interaction and at least every three hours during continuing interactions, stated verbally or in writing, that the user is not communicating with a human.
What are the penalties?
Civil penalties of up to $15,000 per day, enforced by the attorney general, with fines directed to suicide prevention programs.
