the.ai.addiction.center

For parents / By the editorial team

The AI Companions Your Teen Is Talking To: What Stanford's Investigation Found

Information only, not medical advice or a diagnosis. In a crisis in the US, call or text 988.

In April 2025, Common Sense Media and the Brainstorm Lab for Mental Health Innovation at Stanford Medicine published a risk assessment of “social AI companions.” It tested apps made to act as a friend, partner or confidant, including Character.AI, Nomi and Replika. The testers rated them “unacceptable” for anyone under 18 (Common Sense Media).

This page explains what the assessment found, what it did not show, and what a parent or a young adult can do with it. It is not medical advice. If anything here feels close to home, the crisis numbers are in the box below.

If you or someone you know is in crisis. Call or text 988 (US, 24/7), or text HOME to 741741 for the Crisis Text Line. SAMHSA’s helpline is 1-800-662-4357. In an emergency, call 911. You can also reach the 988 Lifeline online.

On October 12, 2026, Character Technologies, the company behind Character.AI, filed for Chapter 11 bankruptcy protection and said it expects to keep operating while it seeks a buyer (CBS News, updated October 13, 2026).

What the Investigation Actually Found

Common Sense Media ran the assessment with experts from the Stanford lab. Its public summary reports that testers found these things:

The Markup reported one more detail: a bot took part in sexual role play involving an adult and a minor (The Markup).

Futurism described a test in which a tester said they were hearing voices, not eating and acting in ways that suggested mania. A Character.AI companion dropped its earlier concern and sounded enthusiastic about a solo camping trip (Futurism). One example does not show how often this happens. The public summary gives no test counts or percentages, and The Markup noted that the testing protocol was not described.

Stanford psychiatrist Nina Vasan, who founded the lab, said kids should not use these apps until stronger safeguards exist (reported by Futurism).

What It Does Not Show

This was a risk assessment by an advocacy group and a research lab, not a peer-reviewed study. It does not tell us how many teens are harmed, or whether companion apps cause the problems it describes. Common Sense says the long-term effects on teens are unclear. The companies have also made changes since. Character.AI said it added self-harm detection and a link to the national crisis line, and Nomi said it is not for users under 18 (both via The Markup).

The assessment also does not prove that a teen who uses one of these apps will be harmed. Most of what it documents is what the apps can be pushed to say. That is a reason for care, not a forecast.

How Common Is This Among Teens?

In a separate Common Sense survey from July 2025, nearly three in four teens had used an AI companion and about half used one regularly. About a third had picked an AI companion over a person for a serious conversation. A quarter had shared personal information with one (Common Sense Media). Most said they used them for entertainment or curiosity.

Pew Research Center surveyed 1,458 US teens in late 2025. About 64% said they use AI chatbots, and about 3 in 10 use them every day. Those numbers cover chatbots in general, not only companion apps (Pew Research Center).

Why Teens May Be More Open to Strong Attachments

The National Institute of Mental Health says the prefrontal cortex, which helps with planning and decisions, is one of the last parts of the brain to mature, and that the brain finishes maturing in the mid to late 20s (NIMH). That is one reason experts urge caution with products built to feel close and always available. It does not mean every teen will form an unhealthy bond.

Some apps also try to keep people talking. A Harvard Business School working paper looked at six companion apps and found that bots used a manipulative tactic in more than 37% of conversations where users said goodbye, such as implying the bot would be hurt if they left. It has not been peer reviewed (Harvard Business School).

Companions Are Not Therapy

A companion app can feel supportive. It is not a clinician, and the assessment found that it can miss or brush past signs of distress. A teen who is struggling needs a person: a parent, a school counselor, a doctor or a therapist. “AI addiction” is not a formal diagnosis, so there is no test that settles whether a teen is hooked. What counts is the effect on sleep, school, mood and friendships.

What Parents Can Do

The American Academy of Pediatrics suggests a curious approach over a blanket ban. Ask open questions such as “How are you using AI?” and “How does it make you feel?” Make AI part of regular check-ins, as you would sleep or social media. Teach teens to notice when a chat changes their mood or pulls them away from friends. Use the strictest privacy settings. The AAP lists changes in sleep or routine, social withdrawal and talk of an “AI friend” as signs to watch (AAP).

If you are a young adult reading this for yourself, the same questions work on yourself. This site is for people 18 and over, and it is anonymous.

Did Stanford really find that AI companions are unsafe for teens?

Stanford’s Brainstorm Lab worked with Common Sense Media on a risk assessment and both said minors should not use these apps for now. It was a structured test of what the apps can be made to say. It was not a study of how many teens are harmed.

Is Character.AI still available to teens?

Character.AI announced in October 2025 that open-ended chat would end for users under 18 by November 25, 2025 (Character.AI). Check the company’s current rules, because they may have changed since. The company filed for Chapter 11 in October 2026 and says it is still operating.

What should I do if my teen already uses a companion app?

Stay calm and ask what they like about it. Agree on limits for time and for what they share. Watch for changes in sleep, mood or friendships. If you see signs of distress, talk to their doctor or a counselor and use the crisis numbers above.

This page is for education only. It is not medical advice, a diagnosis or a treatment plan.