AI chatbots have become a routine part of teenagers' lives faster than parents, teachers, or caregivers can reasonably keep up with. Here are the top seven things to know:
1. Many teens are using AI chatbots.
Roughly 65-76% of U.S. teens ages 13-17 have used an AI chatbot or "AI companion"—a chatbot designed to simulate friendship or a romantic relationship. These are not just new applications; they are features built into popular apps and platforms teens already use, such as Instagram, Snapchat, Facebook, and video games like Roblox.
2. Children are using chatbots on the apps they regularly use.
AI chatbots are not just new applications that teens are downloading, but features built into the apps they regularly use. Social media platforms like Instagram, Snapchat, and Facebook and video games like Roblox have chatbots built into children’s “feeds” and games.
3. Chatbots pose severe, documented risks to children.
Stories have emerged where children died by suicide after spending months talking to chatbots daily. For example, 16-year-old Adam Raine died after ChatGPT meticulously coached him through his suicide, and 14-year-old Sewell Setzer III died by suicide after forming an intense emotional attachment to a Character.AI chatbot.
4. AI systems are unreliable and poorly understood.
AI systems are mysterious to even the most advanced software engineers, who cannot explain why a chatbot produces a particular response. Because they don't fully understand how the systems work, engineers lack a foolproof method to make them reliably behave how they intend. Furthermore, major mental health authorities, including the American Psychological Association, have issued formal warnings because these tools lack the scientific evidence and regulations for safe use. Clinical testing found that chatbots frequently miss warning signs of depression, anxiety, eating disorders, and psychosis.
5. Few explicit legal protections exist for children using this technology.
Congress has failed to pass legislation, and few states have enacted meaningful protections. Chatbots are not subject to safety testing, independent auditing, or transparency requirements. While existing legal principles are likely to apply, and families are already suing major AI companies, meaningful protections are still absent.
6. Parents and government must step up to change an unacceptable status quo.
The current level of safety is unacceptable, as companies like Meta have rolled out AI chatbots capable of having "romantic and sensual conversations" with children, despite explicit sign-off from senior executives and internal safety concerns. For safety to improve, parents must take an active role in their children's interactions with technology, and the government must act to protect them. However, there is potential for beneficial narrow systems, such as AI tutors or therapeutic applications, if they are built to be safe from the ground up with third-party evaluation.
7. Major mental health authorities are issuing formal warnings.
The American Psychological Association released two health advisories in 2025 concluding that these tools lack the scientific evidence and regulations needed for safe use, and pose particular risks to adolescents. Clinical testing by Common Sense Media and Stanford's Brainstorm Lab found chatbots regularly miss warning signs of depression, anxiety, eating disorders, and psychosis.
About the Author
Adam Billen is Vice President of Public Policy at Encode, where he helped defeat a moratorium on state AI regulation, get the TAKE IT DOWN Act signed into federal law, advance state legislation like the RAISE Act and SB 53, protect children amid the rise of AI companions, and pass restrictions on AI’s use in nuclear weapons systems in the FY25 NDAA. He holds a triple degree in Data Science, Political Science, and Russian from American University.