What Teens Are Really Doing with AI Chatbots in 2026

For the first generation growing up with AI, the most popular use is not homework help. It is conversation. And the research on what that means is only beginning to catch up.

When most adults think about teenagers using AI, they picture students cheating on essays or getting homework answers. The reality of how Gen Z actually uses AI is more complicated — and more human.

A New York Times investigation in April 2026 documented a widespread pattern: teenagers using AI companion apps for hours per day, building elaborate roleplay scenarios, treating AI personas as social relationships, and in some cases confiding things in AI chatbots that they would not tell their parents or friends.

The apps they are actually using

Character.AI remains the market leader, with users creating and conversing with custom AI personas — fictional characters, historical figures, celebrities, or original creations. Its user base skews younger than any other major AI platform, with a significant share of users under 18.

Talkie (referenced in the NYT report) runs aggressive social media advertising targeting teens. Its ads feature AI personas with names and personality descriptions — effectively positioning AI as a friend waiting to be met.

Replika was originally designed as an emotional support tool for adults but is widely used by teenagers, particularly those who describe themselves as socially anxious or isolated. Users typically develop a named AI companion over months.

Snapchat My AI introduced millions of younger users to AI conversation through an interface embedded in a social app they already used. Because it appears in the normal Snapchat chat interface, many younger users treat it as a contact.

Why teens prefer it

Interviews with teenagers consistently surface the same themes: AI is available instantly, never judges them, never shares their secrets, and can be anything they need it to be. For a generation navigating complex social hierarchies, academic pressure, and the permanent public record of social media, a private conversation partner with no social stakes has obvious appeal.

Researchers note that some of what teens do with AI companions — practicing conversations, working through emotions, exploring identity through roleplay — are healthy developmental activities. The same things happen in private journaling, with imaginary friends, or in therapeutic fiction. The AI version is simply more responsive.

The concern is not that these activities are inherently harmful, but that the scale and intensity of AI engagement — and the commercial design of these apps to maximize session time — may push usage into territory that real-world relationships and developmental experiences used to occupy.

What the research says

The honest answer is that rigorous long-term research on AI's effects on adolescent development does not yet exist — the tools are too new. What exists is a growing body of observational studies, clinical reports, and smaller controlled experiments.

Preliminary findings suggest:

• Potential benefits: Teens with social anxiety report using AI to practice difficult conversations before having them in real life. Some therapists report patients using AI to process emotions between sessions.
• Identified risks: Heavy users show patterns of preference for AI over peer interaction. In some cases, teens report feeling that AI understands them better than humans — which correlates with reduced investment in human relationships.
• The blurred line problem: Younger teens (under 14) are more likely to misunderstand AI relationships as reciprocal. They may genuinely believe the AI cares about them, misses them, or is hurt if they stop using it.

The regulatory response

China moved in April 2026 to regulate AI companion apps, banning addictive or manipulative AI services targeting children under 18. The regulations specifically prohibit AI apps from using variable reward mechanics — the same psychology behind social media engagement loops — in products accessible to minors.

In the US, the picture is more fragmented. Several states have proposed age verification requirements for AI companion apps. Character.AI settled a federal lawsuit in early 2026 related to content safety for minors and committed to implementing additional safeguards. Federal legislation remains stalled.

Anthropic, OpenAI, and Google have all implemented stronger age gates and content restrictions for minors — but the companion-app category sits in a regulatory grey zone where neither social media law nor AI safety guidelines fully apply.

What parents can actually do

Child psychologists and researchers broadly agree that banning AI apps is both impractical and counterproductive for most families. Teenagers who want to access these tools will find ways to do so; prohibition without understanding closes the conversation.

What tends to work better:

The bigger picture

Every generation has had its technology panic: television, video games, social media. The concerns about AI companions for teenagers are not entirely different in structure — something new and absorbing that displaces time from traditional developmental activities, with uncertain long-term effects.

What is different is the personalization and interactivity. Television is passive. Social media is somewhat interactive. AI adapts to you specifically, remembers what you said yesterday, and responds in ways calibrated to keep you engaged. That combination is new, and the research is genuinely still catching up.

The instinct to either panic or dismiss is understandable. Neither serves teenagers particularly well. What does: treating AI literacy as a basic skill, having honest conversations about how these tools work, and watching for behavioral changes — the same parenting instincts that apply to any influential technology.

Frequently Asked Questions