AI Chatbot Companions: The Emerging Regulatory Framework for Protecting Children and Teens
Laura J. Protzmann, Attorney, National Advertising Division & AI, BBB National Programs; Project Lead, AI Chatbot Accountability Initiative, Center for Industry Self-Regulation
Legislators, regulators, and the public are increasingly expecting AI chatbot platforms to prioritize the safety, privacy, and well-being of users, particularly children and teens. As emotionally responsive AI systems become more sophisticated and widely adopted, lawmakers are focusing not only on privacy and content concerns, but also on the psychological and behavioral impacts of AI chatbot interactions.
In response to the rapid rise of generative AI, even since last year, states including California, Connecticut, Georgia, Idaho, Iowa, Nebraska, New York, Oregon, and Washington have advanced legislation aimed at regulating “AI companion chatbots” and other conversational AI systems that may build human-like emotional engagement with users, rather than simply provide customer service or informational support.
While the bills vary in scope and enforcement, several common themes are emerging:
Many of these proposals reflect long-standing principles in child-directed advertising and digital privacy regulation, including transparency, protections for vulnerable audiences, and limits on manipulative engagement tactics. The policy debate around AI chatbots also cannot occur in isolation. Even when designed for general audiences, these systems inevitably attract younger users.
At the same time, emerging AI chatbot legislation, age-appropriate design codes (AADC), and age verification proposals are beginning to converge into a broader regulatory framework focused on protecting children online.
As lawmakers increasingly expect platforms to safely design products for minors, companies face growing tension between knowing the age of users and minimizing the collection of personal data. These issues are not new. BBB National Programs has decades of experience navigating similar questions at the intersection of children’s privacy, advertising, and emerging technologies.
As AI systems continue to blur the lines between entertainment, companionship, and influence, companies developing conversational AI products for younger users may need to adapt compliance approaches traditionally associated with children’s advertising, online safety, and youth-facing digital products to address new and evolving risks.
Washington HB 2225, effective January 1, 2027, similarly mandates disclosures at the beginning of interactions and every three hours thereafter (every hour for minors), and includes a requirement that chatbots be prevented from generating outputs that contradict these disclosures. Georgia SB 540, effective July 1, 2027, has the same requirement.
These requirements reflect growing legislative concern that emotionally sophisticated AI systems can blur the line between human and machine interaction. As detailed below, several proposals also reflect concern over anthropomorphic AI design, including systems that simulate empathy, companionship, romance, or therapeutic relationships in ways that may encourage users, particularly minors, to form emotional attachments to AI systems.
Connecticut SB 5, Oregon SB 1546, and Washington HB 2225 require systems to also implement protocols for detecting and addressing suicidal ideation, as well as preventing the generation of harmful content encouraging suicide or self-harm. Georgia SB 540 has similar provisions, but with a broader focus on severe harm or related emotional crises.
These provisions reflect increasing concern among lawmakers that emotionally persuasive AI systems may intensify loneliness, dependency, or mental health crises, particularly among younger or otherwise vulnerable users.
Oregon SB 1546, Iowa SF 2417, Idaho SB 1297, and Nebraska LB 525 are narrower but contain some of the same provisions. In addition, along with Georgia SB 540, these laws also contain provisions related to preventing a system of rewards and affirmation designed to maximize engagement.
Proposed legislation in California (SB 1119) and New York (S 9051) also seeks to expand current laws to address these issues.
Together, these bills represent one of the first attempts in U.S. law to directly regulate emotionally manipulative design practices in artificial intelligence systems. Notably, several proposals move beyond traditional content moderation frameworks and instead attempt to regulate the behavioral architecture of AI systems themselves, including features designed to foster emotional attachment, prolong engagement, simulate intimacy, or discourage disengagement.
Taken together, these bills suggest that states are moving quickly to fill a regulatory vacuum around emotionally interactive AI systems.
The bipartisan CHATBOT Act focuses on parental oversight and youth protections for AI chatbot use. The bill would require AI chatbot providers to create family accounts for children under 13 and obtain verifiable parental consent before teens ages 13–17 may access AI chatbots. It also mandates extensive parental controls, including limits on screen time, disabling notifications and rewards systems, monitoring conversation history, and controlling how long chatbot memory retains user data. The legislation prohibits targeted advertising using minors’ personal data and requires clear disclosures that users are interacting with artificial intelligence rather than humans.
The GUARD Act, as amended, takes a more restrictive approach. It would require robust age verification measures for all chatbot users and prohibit minors entirely from accessing “AI companions,” defined to include AI chatbots that simulate a sustained interpersonal relationship or emotional interaction with persistent responses suggesting affection or attachment. The bill also creates criminal penalties for AI chatbots that encourage self-harm, suicide, or violence involving minors. Additionally, it requires disclosures that the chatbot is not human.
Even absent comprehensive federal AI legislation, companies may still face exposure under existing federal consumer protection frameworks. Regulators, including the Federal Trade Commission, have increasingly focused on deceptive design practices, dark patterns, and harms to minors in digital environments, areas that may overlap significantly with emotionally interactive AI systems.
Companies developing conversational AI systems, particularly those accessible to minors, should expect increasing scrutiny around:
The emerging regulatory landscape signals that lawmakers increasingly view emotionally interactive AI systems as a distinct category of higher-risk technology. As state requirements continue to diverge, companies operating nationally may face increasing pressure to adopt uniform youth-safety and transparency standards rather than attempting to tailor conversational AI experiences state-by-state. BBB National Programs is well-positioned to support organizations that seek to proactively implement responsible design and youth-protection frameworks as regulatory expectations continue to evolve.
Legislators, regulators, and the public are increasingly expecting AI chatbot platforms to prioritize the safety, privacy, and well-being of users, particularly children and teens. As emotionally responsive AI systems become more sophisticated and widely adopted, lawmakers are focusing not only on privacy and content concerns, but also on the psychological and behavioral impacts of AI chatbot interactions.
In response to the rapid rise of generative AI, even since last year, states including California, Connecticut, Georgia, Idaho, Iowa, Nebraska, New York, Oregon, and Washington have advanced legislation aimed at regulating “AI companion chatbots” and other conversational AI systems that may build human-like emotional engagement with users, rather than simply provide customer service or informational support.
While the bills vary in scope and enforcement, several common themes are emerging:
- Transparency
- Child safety
- Suicide and self-harm protocols
- Limits on emotional manipulation, and
- Restrictions on anthropomorphic features
Many of these proposals reflect long-standing principles in child-directed advertising and digital privacy regulation, including transparency, protections for vulnerable audiences, and limits on manipulative engagement tactics. The policy debate around AI chatbots also cannot occur in isolation. Even when designed for general audiences, these systems inevitably attract younger users.
At the same time, emerging AI chatbot legislation, age-appropriate design codes (AADC), and age verification proposals are beginning to converge into a broader regulatory framework focused on protecting children online.
As lawmakers increasingly expect platforms to safely design products for minors, companies face growing tension between knowing the age of users and minimizing the collection of personal data. These issues are not new. BBB National Programs has decades of experience navigating similar questions at the intersection of children’s privacy, advertising, and emerging technologies.
As AI systems continue to blur the lines between entertainment, companionship, and influence, companies developing conversational AI products for younger users may need to adapt compliance approaches traditionally associated with children’s advertising, online safety, and youth-facing digital products to address new and evolving risks.
Transparency Disclosures
One of the clearest trends across the legislation is the requirement that AI systems disclose they are not human. California SB 243, effective as of January 1, 2026, requires companion chatbots to provide clear and conspicuous notifications to any user the operator knows is a minor that the user is interacting with artificial intelligence rather than a person, and repeat that notification every three hours along with a reminder to take a break. Iowa SF 2417 and Idaho S 1297, both effective July 1, 2027, and Nebraska LB 525 have similar requirements, with the option of providing a persistent disclaimer.Washington HB 2225, effective January 1, 2027, similarly mandates disclosures at the beginning of interactions and every three hours thereafter (every hour for minors), and includes a requirement that chatbots be prevented from generating outputs that contradict these disclosures. Georgia SB 540, effective July 1, 2027, has the same requirement.
These requirements reflect growing legislative concern that emotionally sophisticated AI systems can blur the line between human and machine interaction. As detailed below, several proposals also reflect concern over anthropomorphic AI design, including systems that simulate empathy, companionship, romance, or therapeutic relationships in ways that may encourage users, particularly minors, to form emotional attachments to AI systems.
Suicide and Self-Harm Protocols
Many of the bills require operators to adopt protocols for detecting suicidal ideation and referring users to crisis resources such as the 988 Suicide and Crisis Lifeline. California SB 243 requires operators to maintain suicide-prevention protocols, including crisis referrals, provide public notice on their website, and report annually to the Office of Suicide Prevention. New York’s Article 47 has similar provisions.Connecticut SB 5, Oregon SB 1546, and Washington HB 2225 require systems to also implement protocols for detecting and addressing suicidal ideation, as well as preventing the generation of harmful content encouraging suicide or self-harm. Georgia SB 540 has similar provisions, but with a broader focus on severe harm or related emotional crises.
These provisions reflect increasing concern among lawmakers that emotionally persuasive AI systems may intensify loneliness, dependency, or mental health crises, particularly among younger or otherwise vulnerable users.
Emotional Dependence and Manipulative Engagement
Nearly every state bill imposes heightened safeguards when minors interact with AI companions. Connecticut SB 5, effective January 1, 2027, and Washington SB 2225 both have extensive requirements related to manipulative techniques that could lead to emotional dependence. Those requirements include restrictions on responses that isolate users from friends and family and manipulative techniques that are intended to extend interaction, including reminding users that the companion can be used for emotional support, excessive praise, mimicking a romantic relationship, and simulating emotional distress, loneliness, guilt, or abandonment if the user seeks to end a conversation or reduce usage. Georgia SB 540 has similar provisions.Oregon SB 1546, Iowa SF 2417, Idaho SB 1297, and Nebraska LB 525 are narrower but contain some of the same provisions. In addition, along with Georgia SB 540, these laws also contain provisions related to preventing a system of rewards and affirmation designed to maximize engagement.
Proposed legislation in California (SB 1119) and New York (S 9051) also seeks to expand current laws to address these issues.
Together, these bills represent one of the first attempts in U.S. law to directly regulate emotionally manipulative design practices in artificial intelligence systems. Notably, several proposals move beyond traditional content moderation frameworks and instead attempt to regulate the behavioral architecture of AI systems themselves, including features designed to foster emotional attachment, prolong engagement, simulate intimacy, or discourage disengagement.
Enforcement
Enforcement mechanisms vary significantly across states. Some bills rely primarily on attorneys general and general consumer protection statutes, while others provide private rights of action. Oregon’s SB 1546 is notable for allowing injured individuals to sue operators directly for actual damages or statutory damages of $1,000 for each violation and injunctive relief. Georgia authorizes civil penalties of up to $10,000 per knowing violation. Washington explicitly classifies violations as unfair or deceptive acts under the state’s consumer protection law.Taken together, these bills suggest that states are moving quickly to fill a regulatory vacuum around emotionally interactive AI systems.
Federal Proposals
While no federal legislation has been enacted yet, several bills have been introduced. For example, the proposed federal CHATBOT Act and GUARD Act both seek to address growing concerns about the impact of AI chatbots on minors, but they take different regulatory approaches.The bipartisan CHATBOT Act focuses on parental oversight and youth protections for AI chatbot use. The bill would require AI chatbot providers to create family accounts for children under 13 and obtain verifiable parental consent before teens ages 13–17 may access AI chatbots. It also mandates extensive parental controls, including limits on screen time, disabling notifications and rewards systems, monitoring conversation history, and controlling how long chatbot memory retains user data. The legislation prohibits targeted advertising using minors’ personal data and requires clear disclosures that users are interacting with artificial intelligence rather than humans.
The GUARD Act, as amended, takes a more restrictive approach. It would require robust age verification measures for all chatbot users and prohibit minors entirely from accessing “AI companions,” defined to include AI chatbots that simulate a sustained interpersonal relationship or emotional interaction with persistent responses suggesting affection or attachment. The bill also creates criminal penalties for AI chatbots that encourage self-harm, suicide, or violence involving minors. Additionally, it requires disclosures that the chatbot is not human.
Even absent comprehensive federal AI legislation, companies may still face exposure under existing federal consumer protection frameworks. Regulators, including the Federal Trade Commission, have increasingly focused on deceptive design practices, dark patterns, and harms to minors in digital environments, areas that may overlap significantly with emotionally interactive AI systems.
What Companies Should Be Preparing For
Although Congress has yet to enact comprehensive AI legislation, the emerging legal consensus is that AI systems that simulate human relationships, especially for children or vulnerable users, will face heightened scrutiny, mandatory transparency requirements, and growing obligations to mitigate emotional and psychological harm.Companies developing conversational AI systems, particularly those accessible to minors, should expect increasing scrutiny around:
- Whether users understand they are interacting with AI
- How systems respond to mental health crises or self-harm discussions
- Whether engagement features encourage emotional dependency
- The use of persuasive or anthropomorphic design patterns
- Age assurance and parental control mechanisms
- Internal governance, testing, and documentation of youth-safety measures
The emerging regulatory landscape signals that lawmakers increasingly view emotionally interactive AI systems as a distinct category of higher-risk technology. As state requirements continue to diverge, companies operating nationally may face increasing pressure to adopt uniform youth-safety and transparency standards rather than attempting to tailor conversational AI experiences state-by-state. BBB National Programs is well-positioned to support organizations that seek to proactively implement responsible design and youth-protection frameworks as regulatory expectations continue to evolve.