Consumer Protection - Companion Chatbots - Regulation
HB952 creates a new regulatory framework for “companion chatbots,” defined as AI systems with natural-language interfaces that provide adaptive, human-like responses and are capable of sustaining a relationship across multiple interactions. The bill requires operators of these chatbots to adopt and publish protocols aimed at preventing harmful outputs, especially content related to self-harm, suicidal ideation, suicide, and sexually explicit conduct. It also requires clear warnings that the chatbot is artificially generated and not human, with additional recurring warnings for minor users during extended use.
The bill further limits data collection and use by operators, including restricting the collection of personal data to what is reasonably necessary and prohibiting the use of emotional-state or mental-health vulnerability data to increase chatbot engagement. Operators must also maintain a complaint system, investigate complaints within three days, and report complaint outcomes and protocol information to the Maryland Office of Suicide Prevention. The Office must compile and publish annual data from operator reports beginning in 2027. Finally, the bill treats violations as unfair, abusive, or deceptive trade practices under the Maryland Consumer Protection Act and also provides that chatbots are considered products for certain product-liability claims, including design defect, manufacturing defect, and marketing defect actions.
HB952 amends the Commercial Law Article by adding a new section regulating companion chatbots and by expanding the list of conduct treated as an unfair, abusive, or deceptive trade practice. It creates new compliance duties for operators and developers, imposes reporting obligations to the Office of Suicide Prevention, and authorizes consumer-protection enforcement and penalties for violations. The bill also changes liability exposure by expressly classifying chatbots as products for purposes of product-liability actions, potentially affecting operators, developers, and other entities involved in chatbot deployment in Maryland.
The bill appears to have broad support in the House, passing third reading 119-3, and the committee report was favorable with amendments. That vote margin suggests strong bipartisan concern about the risks posed by AI companion systems, particularly for minors and vulnerable users. The absence of recorded committee transcripts limits insight into detailed debate, but the overall legislative posture indicates general agreement on the need for consumer protections and safety guardrails.
The main points of contention likely center on how far the state should go in regulating AI chatbots and assigning liability. The bill imposes significant operational duties, including content filtering, warning requirements, reporting, and complaint handling, which may concern chatbot operators and developers as burdensome or difficult to implement. Another likely issue is the product-liability provision, which could expose companies to strict liability and defect claims for harms caused by chatbot interactions. Privacy and data-use restrictions may also be debated, especially the prohibition on using emotional-state or mental-health vulnerability data to optimize engagement.