S. 896 would create a new chapter in Title 39 of the South Carolina Code regulating chatbots and other AI-driven conversational systems. The bill, as amended by committee, is titled the “Protecting Children from Chatbots Act” and applies to operators of chatbots with at least 50,000 monthly active users worldwide, while excluding certain internal tools, customer-service bots, government service bots, and limited video game chatbots. It defines key terms such as chatbot, chat log, personal data, restricted feature, emotional dependence, covered harm, and reasonable age verification.
The bill requires covered entities to offer a limited-access mode for unverified users and to restrict minors from accessing certain features unless parental consent is obtained. Restricted features include personalization, proactive outreach, long or extended sessions, relationship simulation, and explicit content. For minors with parental consent, the bill requires parental controls, optional parental access to chat logs, and immediate parental notice in certain crisis situations. It also limits how chat logs and personal data of minors may be used, including prohibiting use for advertising and generally prohibiting model training on minor data except for safety, compliance, or harm-prevention purposes.
The bill also imposes broader safety, transparency, and data-handling duties on covered chatbot operators. These include prohibitions on making chatbots appear to be human or licensed professionals, claiming chat logs are confidential, selling chat logs, or retaining chat logs for more than ten years absent legal necessity. Operators must maintain a written data security program, take steps to prevent re-identification of de-identified data, identify emotional dependence and suicidal or self-harm crises, and provide crisis messages and, where possible, notify emergency services and parents. The Attorney General is authorized to adopt implementing rules and regulations and to receive reports of covered incidents involving death, suicide attempts, self-harm, psychiatric emergencies, or serious physical injury.
The bill would significantly affect state law by adding a new regulatory and enforcement framework for chatbot providers, including private rights of action, state enforcement authority, civil penalties, and limits on contractual waivers and arbitration. It also treats chatbots as products for product-liability purposes and creates duties intended to reduce harm to users, especially minors. The measure would apply prospectively upon gubernatorial approval, and existing accounts would need age classification or restricted-feature disablement within 60 days.
Overall sentiment appears generally supportive of stronger chatbot safeguards, especially for children and vulnerable users, as reflected by the committee’s recommendation that the bill do pass with amendment. The main points of contention likely center on the breadth of the regulation, the compliance burden on AI providers, the scope of age verification and parental access requirements, and the bill’s treatment of data use, model training, and liability. No recorded floor votes or committee debate transcripts were provided, so the available record shows committee approval but not detailed opposition or support arguments.
The bill would add a new chapter to Title 39 regulating chatbot operators in South Carolina, imposing age-verification, parental-consent, content, data-use, safety, reporting, and disclosure requirements on covered entities. It creates enforcement authority for the Attorney General and private civil actions, authorizes penalties and damages, limits arbitration and waiver clauses, and establishes chatbot-related duties that operate alongside existing law. It would also require certain operators to change product design and data practices, especially regarding minors, chat logs, and crisis-response procedures.
The available record suggests a favorable committee posture toward the bill, since the Senate Labor, Commerce and Industry Committee recommended that it pass with amendment. The bill’s framing around child protection, crisis response, and transparency indicates a policy goal likely to attract support from lawmakers concerned about AI harms. No vote tally or transcript is available, so there is no documented floor-level opposition or debate in the provided materials.
Likely areas of contention include whether the bill’s age-verification and parental-consent requirements are practical or overly burdensome, whether the restrictions on training and data use for minors are too broad, and whether the liability and private-right-of-action provisions expose providers to significant litigation risk. Another likely dispute is the scope of the prohibited chatbot features, especially relationship simulation, proactive outreach, and long-context interactions, as well as whether the bill could chill beneficial AI uses in education, customer service, or mental-health-adjacent settings. The committee amendment suggests the bill may have been refined to address some of these concerns, but no transcript is available to identify specific objections or sponsors’ responses.