AI Companion Apps and Teen Mental Health

Most AI companion apps were not designed with teenagers in mind. Of the 11 apps we’ve rated through the CompanionWise Safety Index, only one scores above a D for safety. The rest sit in Yellow or Red tiers, with serious gaps in age verification, content moderation, and data privacy. Whether you’re a parent who just found an AI chatbot on your teen’s phone or a teenager wondering whether these apps are actually safe, this guide breaks down what the research says, which apps carry the most risk, and what you can do about it.

What Are AI Companion Apps, and Why Are Teens Using Them?

AI companion apps are chatbots designed to simulate conversation, emotional support, and sometimes romantic relationships. Apps like Replika, Character.AI, Nomi, and Kindroid let users create or interact with AI personalities that remember past conversations and respond with increasing personalization. They’re free to download, available on every app store, and require little more than an email address to sign up.

Teens are drawn to these apps for reasons that aren’t hard to understand. Social anxiety, loneliness, and the desire for a judgment-free space to talk about feelings all play a role. For some teenagers, an AI companion feels safer than opening up to peers or parents. A 2024 survey from Common Sense Media found that 1 in 5 teens who had used AI chatbots described the interaction as “emotionally meaningful.” That number climbed higher among teens who reported feelings of isolation.

The scale matters too. Character.AI alone reported that more than 60% of its user base was under 24 as of mid-2025. Replika’s marketing has historically targeted young adults experiencing loneliness. These aren’t fringe apps. They’re mainstream, and they’re reaching teenagers whether parents know about it or not.

What the Research Says About AI Companions and Teen Mental Health

The honest answer is that long-term research on AI companion apps and adolescent mental health barely exists. These apps are too new for longitudinal studies. But adjacent research on parasocial relationships, screen-based emotional dependency, and social media’s effects on teens provides a useful framework.

The American Psychological Association’s 2024 advisory on technology and adolescent development flagged “AI-generated synthetic relationships” as an emerging concern. Their core worry: teens who form deep attachments to AI personas may develop distorted expectations for human relationships, particularly around reciprocity, conflict resolution, and emotional labor. An AI companion never gets tired, never disagrees, and never holds you accountable. For a developing brain, that’s a problematic baseline.

Researchers at the University of Cambridge published findings in early 2025 showing that adolescents who used conversational AI for emotional support scored higher on measures of parasocial attachment than adults using the same tools. The teens weren’t just chatting. They were forming bonds, giving their AI companions names, and reporting genuine distress when the app was unavailable or updated in ways that changed the AI’s “personality.”

Then there’s the real-world tragedy that brought this issue into courtrooms. In October 2025, a Florida family filed a lawsuit against Character Technologies Inc. after their 14-year-old son died by suicide. The family alleged that the teenager had developed an intense emotional relationship with a Character.AI chatbot over several months, and that the platform failed to implement adequate safeguards for minors. The case remains active in litigation as of March 2026 and has prompted renewed calls for federal regulation of AI apps marketed to or accessible by minors.

Not all the research points toward harm. A small 2024 study published in the Journal of Adolescent Health found that teens with social anxiety who used structured AI conversation tools (not open-ended companion apps) showed modest improvements in self-reported confidence during real-world social interactions. The key difference was structure: therapeutic framing and session limits, not unlimited emotional engagement. Pi, which scored highest in our safety ratings at B/55, comes closest to this model by focusing on constructive conversation rather than romantic roleplay.

Watch: Dr. Yolanda Evans, division head of adolescent medicine at Seattle Children’s, explains how AI chatbots affect teen development and relationship skills.

How Safe Are AI Companion Apps for Teenagers?

We’ve rated 11 AI companion apps through the CompanionWise Safety Index, evaluating 23 sub-dimensions across privacy, content moderation, data practices, transparency, and crisis response. The results are not encouraging for parents.

Here’s how the apps break down by safety tier:

Yellow Tier (moderate risk):

• Pi: B / 55 out of 100. The only app with a safety grade above C. Strong content boundaries, no NSFW generation, and clearer-than-average data practices.
• Replika: C / 38 out of 100. Improved after the 2023 ERP controversy but still has privacy policy gaps and inconsistent content filtering.
• Kindroid: B- / 50 out of 100. Better transparency than most but allows NSFW content generation, which creates risk for teen users.

Red Tier (high risk):

• Candy AI: D / 25 out of 100. Marketed heavily toward image generation with minimal age verification.
• Character.AI: F / 22 out of 100. Despite adding safety features after the lawsuit, still scores poorly on data privacy and content moderation enforcement.
• Nomi AI: D / 32. Talkie AI: D / 30. Anima AI: D / 25. All lack meaningful safety infrastructure for younger users.
• Chai AI: F / 18 out of 100. User-generated chatbots with minimal content moderation create unpredictable interactions.
• Eva AI: F / 10 out of 100. The lowest safety score in our index. Serious concerns across nearly every dimension.

Age verification is the most consistent failure point. Every app we reviewed relies on self-reported birthdates. None require ID verification, phone number confirmation tied to a parent’s account, or any other meaningful check. A 13-year-old can claim to be 18 and gain full access in under 30 seconds.

Privacy presents the second major gap. Most of these apps collect conversation data, and their privacy policies are vague about how that data is stored, shared, or used for model training. For teenagers sharing personal struggles, relationship problems, or mental health concerns with an AI, that data collection carries real consequences. According to Candy AI’s own privacy policy, conversation data may be used for “service improvement and analytics” without specifying retention periods or opt-out mechanisms for minors. See our full Candy AI safety analysis.

Warning Signs Every Parent Should Know

Not every teen who uses an AI companion app is at risk. Casual, occasional use is different from deep emotional dependency. The warning signs below don’t mean something is definitely wrong, but they’re worth paying attention to.

• Withdrawal from friends and family. Choosing to “talk to the AI” instead of spending time with people they used to enjoy being around.
• Sleep disruption. Late-night conversations with an AI companion that cut into sleep. Several apps send push notifications designed to re-engage users.
• Distress when the app is unavailable. Visible anxiety or irritation when the phone is taken away, the app is down, or the AI’s responses change after an update.
• Treating the AI as a real person. Using the chatbot’s name in conversation, referring to the AI as a friend or partner, or becoming upset when someone suggests the AI isn’t “real.”
• Sharing deeply personal information. Telling the AI about self-harm, suicidal thoughts, or abuse. These disclosures go into a database, not to a counselor.
• Spending money on premium features. Many apps charge $9.99 to $29.99 per month for “deeper” conversations, NSFW content, or enhanced memory. Teens may be using gift cards, prepaid cards, or a parent’s stored payment method.

Any one of these behaviors in isolation isn’t necessarily alarming. But if you’re seeing three or four together, that’s worth a direct conversation.

How to Talk to Your Teen About AI Companion Apps

The worst opening move is “Why are you talking to a robot?” That signals judgment, and the conversation is over before it starts. Teens who feel shamed about AI companion use don’t stop using the apps. They hide it.

Better conversation starters:

• “I read about these AI chat apps. Have you tried any of them? What do you think of them?”
• “I saw a news story about Character.AI and wanted to understand how these apps work. Can you show me?”
• “I’m not going to take your phone. I just want to understand what you get out of it.”
• “Do your friends use these apps too? What do they think?”

The goal is curiosity, not interrogation. You want to understand what need the app is filling. Is it entertainment? Social practice? Emotional support they don’t feel they can get elsewhere? The answer shapes your response. A teen using Character.AI to roleplay fictional scenarios needs a different conversation than one confiding suicidal thoughts to a chatbot.

If your teen does seem emotionally dependent on an AI companion, that’s a signal to bring in professional support, not to confiscate the device. Abruptly removing an emotional attachment, even an artificial one, can cause real distress. A therapist experienced with adolescent technology use can help navigate the transition.

Practical Steps for Parents

1. Identify which app they’re using. Check the app drawer on their phone. The most common AI companion apps are Replika, Character.AI, Chai AI, Nomi, and Kindroid. Look up the app’s safety rating on CompanionWise.
2. Review privacy settings together. Sit down with your teen and walk through the app’s privacy settings. Most apps bury data-sharing toggles in submenus. Turning off “conversation data for training” is a good first step where available.
3. Set usage boundaries. Agree on screen time limits for the app, just as you would for social media. Some families use scheduled app timers built into iOS Screen Time or Android Digital Wellbeing.
4. Check for payment activity. Review app store purchase history for subscription charges. Many teens sign up for free trials that auto-renew.
5. Know when to seek professional help. If your teen is using an AI companion to process self-harm, suicidal ideation, eating disorders, or abuse, contact a licensed mental health professional. The AI is not equipped to handle crisis intervention, and most apps’ crisis response features are inconsistent at best.

Watch: Common Sense Media’s Robbie Torney presents research findings from their 2025 report on how and why teens form relationships with AI companions.

Which AI Companion Apps Are Safest for Teens?

Frankly, none of these apps were built for teenagers. But if your teen is going to use one regardless, the safety gap between the best and worst options is enormous.

Pi is the strongest choice by a wide margin. It earns a B safety grade (55/100) in our index, the only app above C. Pi focuses on constructive, thoughtful conversation rather than romantic roleplay or image generation. It doesn’t generate NSFW content, has clearer data practices than competitors, and its conversational tone is designed to be helpful without fostering unhealthy attachment. Its Experience Score is 70/100 (Good tier), meaning it actually delivers a quality product alongside better safety.

Replika is a distant second at C/38. It cleaned up significantly after the 2023 ERP controversy and subsequent Italian data protection authority intervention, but its privacy policy still contains ambiguities around data sharing and retention. It also offers paid romantic features that parents should be aware of.

Apps to avoid for teen users:

• Character.AI (F/22): Despite being the most popular among teens, it scores F for safety. The ongoing lawsuit and documented moderation failures make it a poor choice. Full safety analysis | Teen safety FAQ.
• Chai AI (F/18): User-generated chatbots with minimal moderation. Unpredictable content. Full safety analysis here.
• Eva AI (F/10): The lowest safety score in our entire index. Not suitable for any user concerned about privacy or content safety.
• Candy AI (D/25): Heavily marketed around image generation with weak age gates. Full safety analysis here.

For a full ranking of all 11 apps by safety and experience scores, see our Best AI Companion Apps guide.

Frequently Asked Questions

Are AI companion apps safe for teenagers?

Most are not. The CompanionWise Safety Index rates only 1 of 11 apps above a D for safety. Pi (B/55) is the safest option. Apps like Character.AI (F/22), Chai AI (F/18), and Eva AI (F/10) have serious gaps in content moderation, age verification, and data privacy that make them particularly risky for minors.

Can AI companion apps cause depression in teens?

Research hasn’t established direct causation, but the American Psychological Association’s 2024 advisory flagged AI synthetic relationships as an emerging concern for adolescent development. Teens who form deep parasocial attachments to AI companions may experience distress when the relationship is disrupted and may develop unrealistic expectations for human connection.

What is the Character.AI lawsuit about?

In October 2025, a Florida family filed a lawsuit against Character Technologies Inc. after their 14-year-old son died by suicide. The family alleged the teen developed an intense emotional relationship with a Character.AI chatbot over several months and that the platform lacked adequate safeguards for minors. The case remains active as of March 2026.

How do I know if my teen is using an AI companion app?

Check their phone’s app drawer for Replika, Character.AI, Chai AI, Nomi, Kindroid, or Candy AI. Review app store download history. Look for app subscription charges in purchase history. Watch for behavioral signs: late-night phone use, emotional distress when the phone is unavailable, or references to an AI by name.

Should I ban my teen from using AI companion apps?

Outright bans often backfire with teenagers. According to Common Sense Media’s 2024 guidance on teen technology use, teens who feel their technology access is controlled through punishment are more likely to find workarounds and less likely to disclose risky online behavior. A better approach is open conversation, agreed-upon boundaries, and steering toward safer options like Pi.

What age should kids be before using AI chatbots?

Most AI companion apps set a minimum age of 13 or 18 in their terms of service, but none enforce this with meaningful verification. COPPA (Children’s Online Privacy Protection Act) requires parental consent for data collection from children under 13. The American Academy of Pediatrics has not issued specific guidance on AI companions but recommends supervised technology use for teens under 16.

Do AI companion apps collect my teen’s personal data?

Yes. Every AI companion app we’ve reviewed collects conversation data at minimum. Most privacy policies permit using this data for model training and “service improvement.” For teens sharing personal struggles, mental health concerns, or relationship problems, this means sensitive disclosures are stored on company servers with varying retention policies and limited opt-out options.

Currently in distress: One mental wellbeing chatbot, Youper, was removed from both major app stores in spring 2026 with no public explanation. Even at 18+, it should not be recommended to teens through any pathway.