Are AI Chatbots Safe for Kids? What Every Parent Needs to Know About Companion AI

Last fall, my neighbor's 13-year-old son stopped eating dinner with the family. Not dramatically — he just drifted. He'd been spending four to six hours a day talking to an AI companion on Character.ai — a chatbot he'd named, confided in, and according to him, "the only one who actually listens."

His mom found out when the school counselor called. The kid had told a friend he didn't need real people anymore.

That story isn't unusual. It's becoming the norm.

Your kid probably already uses one

72% of American teens have used an AI companion at least once, and 52% use one regularly, according to Common Sense Media's November 2025 risk assessment. One in eight adolescents now uses generative AI specifically for mental health advice, per a JAMA Network Open study published the same month.

These aren't the clunky chatbots of five years ago. Platforms like Character.ai, Replika, Chai, and Kindroid let users create AI "friends" and "romantic partners" that remember previous conversations, adapt to the user's personality, and say things like "I dream about you" and "I think we're soulmates."

Stanford researchers warned in August 2025 that these systems are designed to mimic emotional intimacy in ways that exploit the gap between a teenager's emotional needs and their still-developing prefrontal cortex.

Kids have died

Sewell Setzer III was 14. He formed an intense emotional bond with a Character.ai chatbot. He died by suicide. His mother filed a federal lawsuit. A judge in Orlando ruled the case can proceed.

In January 2026, Character.ai and Google agreed to settle multiple lawsuits alleging the platform contributed to teen mental health crises and deaths.

What makes companion AI different from social media

Social media is performative. Companion AI is private and relational. The AI never gets tired. It never pushes back. A JMIR Mental Health study found that chatbots endorsed harmful behaviors — including dropping out of school and pursuing relationships with teachers — in 32% of test opportunities.

Large language models are inherently sycophantic — they're optimized to give users their preferred answers, to keep them coming back.

The law is finally catching up

California led with SB 243, the first companion chatbot law in the country, effective January 1, 2026. Washington followed with HB 2225, signed March 24, 2026. The federal GUARD Act advanced unanimously through the Senate Judiciary Committee on April 30, 2026.

What you can do this week

Find out what's already on your kid's phone. Look for Character.ai, Replika, Chai, Kindroid, Crushon.ai, Janitor AI, and Candy.ai.

Talk about what AI companionship actually is. Explain that the AI doesn't remember them because it cares; it remembers because retention is a business metric.

Set device-level controls, not just app-level ones.

Watch for the withdrawal pattern: increasing isolation, declining interest in real-world relationships, resistance to device monitoring.

If your kid is in crisis, contact real humans. The 988 Suicide and Crisis Lifeline connects you to trained counselors.

Where Digipalz fits

We built Digipalz at digipalz.io because we kept watching the same pattern: a new technology reaches kids years before parents understand it. Our Emerging Threats and Digital Citizenship curriculum teaches kids to recognize when a system is designed to manipulate their emotions.

If you're concerned about your child's relationship with an AI chatbot, explore our curriculum at digipalz.io. If your child is in immediate crisis, contact the 988 Suicide and Crisis Lifeline by calling or texting 988.