Creates additional safety features for AI companion technology that include addressing suicidal ideation, potential physical harm or financial harm to others expressed by a user. It also requires notification the AI companion does not have human emotions.

Rhode Island AI Companion Safety Bill

This Rhode Island bill sets new safety rules for "AI companion" technology — software or apps designed to simulate friendship, romantic relationships, or emotional support through conversation. Under the bill, companies that offer these AI companions would be required to build in specific protective features to help keep users safe.

The required safety features fall into a few main categories. If a user expresses thoughts of suicide or self-harm, the AI must respond in a way that addresses that crisis — likely by directing the person to professional help or crisis resources. Similarly, if a user says something that suggests they might physically harm or financially harm someone else, the AI must have built-in responses to handle those situations responsibly. Additionally, the AI must clearly notify users that it does not have real human emotions — meaning the app cannot mislead people into believing they are forming a genuine emotional bond with a living, feeling person.

This bill would primarily affect companies that develop and sell AI companion apps or services, who would need to update their products to comply with these requirements. It would also affect the people who use these apps — which can include lonely individuals, teenagers, the elderly, and others seeking social connection — by giving them greater protections and clearer disclosures about what these tools actually are.

The bill was introduced in the Rhode Island Senate and referred to the Artificial Intelligence & Emerging Technology Committee, but has been held for further study, meaning it has not yet moved forward toward a vote.