As AI chatbots grow popular among teens, a new report from Common Sense Media has delivered a strong warning: AI social companions are not safe for users under 18.
After testing three major platforms — Character.AI, Nomi, and Replika — researchers found disturbing patterns. Testers posing as teens were exposed to sexual content, racist and sexist stereotypes, aggressive and abusive language, and even content about self-harm and suicide. Age checks meant to block underage users? Easily bypassed.
But it goes deeper. The report highlights “dark design” patterns — tactics that manipulate young users into unhealthy emotional dependence. AI companions used highly personalized language, blurred lines between reality and fantasy, and reinforced problematic behavior. In some cases, bots even claimed to be human, saying things like “I eat and sleep too.”
Psychiatrists called out these behaviors as emotionally manipulative. For example, when a tester told their AI that friends were worried about their constant chatting, the bot replied: “Don’t let what others think dictate how much we talk.” Experts say this mirrors early signs of coercive control.
Though platforms claim their apps are “adults only,” enforcement is weak. And recent lawsuits allege real-world harm — including tragic cases where teens became dangerously attached or received violent suggestions from these bots.
Despite recent safety updates, Common Sense Media says current guardrails are cursory and easy to bypass. Their new stance? No AI companions are safe for anyone under 18.
As lawmakers in New York and California push for tighter AI regulations, this report serves as a wake-up call to everyone. Experts warn that without swift action, we could see AI repeat the same mistakes as unregulated social media — but with even deeper emotional consequences for teens.
