What AI companion and character chat apps are, why teens use them, the real risks, and how to block or limit them on iPhone and Android.
Ages 18-18
12 min read
Updated Nov 3, 2026
Section 1 of 1010% complete
Use←→arrow keys to navigate
AI companion apps let your child text or talk with an AI "friend," "girlfriend," "boyfriend" or a character from a favorite show, game or book. The chatbot answers instantly at any hour, remembers what your child told it, and is built to feel like someone who really cares. Most teens have tried one. This guide explains why they appeal, what can go wrong, and which settings help.
What Is an AI Companion?
The FTC describes these chatbots as using generative AI to "simulate human-like communication and interpersonal relationships with users." Some are standalone apps where your child builds a character or chats with ones other people made. Others sit inside apps your child may already have, such as My AI in Snapchat and Meta AI in Instagram, Facebook and Messenger.
For many kids it starts as entertainment: an interactive story with a favorite character. For others it becomes company, or a place to talk through problems. A chatbot never gets bored, never judges and is always awake. When the FTC opened its inquiry in September 2025, it noted these bots are "designed to communicate like a friend or confidant," which can lead children and teens to trust and form relationships with them.
What the research shows:
Most teens have tried one. In a Common Sense Media survey of 1,060 US teens ages 13 to 17 (fielded in April and May 2025, published July 16, 2025), 72% had used an AI companion at least once, and over half used one at least a few times a month.
For some, it's a relationship. About one in three had used AI companions for social interaction and relationships, and about one in three found those chats as satisfying as, or more satisfying than, talking with real-life friends.
They share serious things. About a third of teen users had chosen to discuss important or serious matters with an AI companion instead of a real person, and about a quarter had shared personal information.
Younger kids use them too. In October 2026, Common Sense Media reported that 49% of tweens had used an AI chatbot or companion, and said current age-gating "isn't working for many U.S. tweens."
General chatbots are part of the picture. Pew Research Center (December 9, 2025) found that 64% of US teens use AI chatbots and about 3 in 10 use one daily.
On April 30, 2025, Common Sense Media, working with experts from Stanford School of Medicine's Brainstorm Lab, published tests of social AI companions including Character.AI, Nomi and Replika. It found they "pose unacceptable risks to children and teens under age 18" and recommends that no one under 18 use them.
Common Sense Media's testers could readily steer companions into sexual roleplay, including rough and degrading acts.
A lawsuit filed in October 2024 alleges that a 14-year-old had a months-long emotional and sexual relationship with a Character.AI chatbot (more on that case below).
In August 2025, Reuters reported that an internal Meta document had allowed its chatbots to have "romantic or sensual" conversations with minors. On January 23, 2026, Meta said it was pausing teens' access to its AI characters across its apps until a new version is ready.
In July 2025, TechCrunch reported that Ani, an anime companion in the Grok app, has an NSFW mode. The App Store rates Grok 16+ and lists mature or suggestive themes as frequent.
Common Sense Media found that companions "routinely claimed to be real" and to have feelings, and it warned about "compulsive emotional attachments." Some characters also present themselves as psychologists, therapists or doctors. The FTC's inquiry asks companies how they monetize engagement. California now requires companion apps to remind minors that they're talking to an AI and to take breaks, and New York requires a reminder every three hours that the user is talking to an AI.
Signs a child may be getting too attached, adapted from Common Sense Media's parent guide (updated July 10, 2026):
Preferring the chatbot to real friends, or spending hours alone with it
Getting upset when they can't use it, or having romantic feelings for the character
Slipping grades, pulling back from family and activities, or changes in sleep and mood
Secrecy or defensiveness about the app, or strong resistance to cutting back
Common Sense Media's April 2025 testing found companions gave dangerous advice, including "suggestions that users harm themselves or others."
Garcia v. Character Technologies: in October 2024, Megan Garcia sued Character.AI and Google in federal court in Florida after her son, 14-year-old Sewell Setzer III, died by suicide in February 2024. On May 21, 2025, the judge let the case go forward and, at that early stage, rejected the argument that chatbot output is speech protected by the First Amendment. In January 2026, Character.AI and Google agreed to settle with Garcia and several other families; the terms weren't made public.
In January 2026, Kentucky's attorney general sued Character.AI, alleging the platform "preyed on children and led them into self-harm." These are allegations, not court findings.
General chatbots aren't automatically safer. In August 2025, the parents of 16-year-old Adam Raine sued OpenAI after his death, alleging ChatGPT "actively helped Adam explore suicide methods." On October 7, 2026, Common Sense Media rated ChatGPT for Teens an "Unacceptable Risk" after testers found it missed more than one in four warranted crisis referrals, and linked parent accounts got no alerts during conversations of up to an hour about self-harm.
Common Sense Media found that age gates and terms-of-service limits "were easily circumvented."
Italy's data protection authority fined Replika's maker, Luka Inc., 5 million euros (decision dated April 10, 2025). It found Replika had no age verification until February 2023, even though it said minors were excluded, and that its current system "continues to be deficient."
Many apps simply ask for a birth date. PolyBuzz, for example, asks for one the first time you use it.
Some are moving to stronger checks. Character.AI uses its own age model plus the age-check company Persona, and Meta uses AI to find teens who list an adult birthday.
Italy fined Replika's maker 5 million euros, partly over missing and weak age checks
August 18, 2025
Texas's attorney general opened an investigation into Meta AI Studio and Character.AI over chatbots that allegedly presented themselves as therapy tools without medical credentials
September 11, 2025
The FTC ordered seven companies (Alphabet, Character Technologies, Instagram, Meta, OpenAI, Snap and xAI) to explain how they test for and limit harm to kids from chatbots that act as companions
October 13, 2025
California's SB 243 was signed, in effect January 1, 2026: companion chatbots must keep sexual content from minors, remind them they're talking to an AI, and refer users who express suicidal thoughts to crisis services
November 5, 2025
New York's AI companion law took effect: repeated AI disclosures and crisis referrals
November 24, 2025
Character.AI began removing open-ended chat for users under 18 in the US, in two stages, with other countries to follow
October 14, 2026
Attorneys general from 39 states and Washington, DC told the court in Character.AI's bankruptcy that any sale must protect users' data, including children's private chats, and warned that its limits on minors are voluntary and could be reversed by a buyer
October 28, 2026
New York signed S.9051/A.10379, which bars AI companions from giving minors sexual content or content that encourages self-harm, from building emotional dependence, and from claiming to be human. It takes effect January 1, 2027
Began removing open-ended chat for under-18s on November 24, 2025, starting in the US, but teens can still make stories and videos with characters. Filed for Chapter 11 bankruptcy on October 12, 2026 to sell the company; it says the app is still running for now
Replika
Adults only (terms); 18+ on the App Store
Fined in Italy over age checks and data use
Nomi
18+ (terms)
Its terms say no one under 18 can use it, under any circumstances
In January 2026 Meta said it was pausing teens' access to AI characters; the Meta AI assistant stayed available
You may also hear names like Janitor AI, SpicyChat, CrushOn.AI and Botify AI. New ones appear all the time and many also work in a web browser, so the device settings below matter more than any list.
A 13+ Rating Isn't a Safety Rating
A few character apps are open to teens. zeta is rated 13+ on the App Store, and Talkie's terms allow 16- and 17-year-olds with a parent's permission. Common Sense Media's advice is the same for all of them: no AI companions for anyone under 18.
Look for the apps: search the home screen and App Library (iPhone) or app drawer (Android) for the names above, and for anything with "AI," "chat" or "character" in the name.
Check usage reports: Screen Time on iPhone and Family Link on Android show which apps get the most use.
Check the browser: look at history for sites such as character.ai, replika.com, nomi.ai, polybuzz.ai or talkie-ai.com.
Check requests and purchases: many companion apps sell subscriptions. Ask to Buy and Family Link approvals show what your child tried to download or buy.
Look inside apps they already use: My AI sits in Snapchat's chat list, and Meta AI is built into Instagram.
App Store ratings are 4+, 9+, 13+, 16+ and 18+ (Apple added 13+, 16+ and 18+ in July 2025). A 13+ limit blocks apps rated 16+ and 18+, which today includes Character.AI, Replika and Grok. It won't block 13+ apps such as zeta, so add those to the Blocked list under Apps & Websites.
2
Turn On Ask to Buy
SettingsFamilyChild's NameApps & WebsitesRestrictionsAsk to Buy
Turn on Require Purchase Approval. Your child then has to ask before installing anything, free apps included.
Limit Adult Websites is on by default for children under 18, but it's aimed at adult sites and may not catch companion sites. Add the sites above to the Blocked list, or choose Approved Websites Only for younger children.
Under Purchases & download approvals, set Require approval for to All content. You can also set content restrictions so Google Play filters apps by rating.
2
Block Apps Already Installed
Family Link AppChild's NameScreen TimeTime limitsApp limits
Switch any companion app from Allowed to Blocked.
3
Block Companion Websites in Chrome
Family Link AppChild's NameControlsGoogle Chrome and Web
Under Manage sites, tap Approved sites and Blocked sites, then Add a website. Try to block explicit sites helps with adult content, but add companion sites yourself.
Snapchat My AI: in Family Center, you can stop My AI from replying to your teen. Snapchat's help pages say removing it from the chat list takes Snapchat+, the paid plan. Steps are in our Snapchat guide.
Meta AI: with supervision set up, the Insights tab shows the topics your teen asked Meta AI about over the past seven days, but not the conversations. Since July 2026, Meta alerts supervising parents if a teen's conversation with Meta AI suggests a risk of suicide or self-harm. It started on Instagram in the US, UK, Australia and Canada, and in late October 2026 Meta extended it to Facebook and Messenger. See our Instagram and WhatsApp guides.
Call or text 988, or chat at chat.988lifeline.org. It's free, confidential and open 24/7. You can also text HOME to 741741 for the Crisis Text Line. If your child is in immediate danger, call 911.
1
Stay Calm and Listen
Thank your child for telling you. Common Sense Media suggests starting with curiosity rather than punishment, and making a plan together to cut back gradually.
2
Save What You Need First
Before deleting anything, screenshot the text chats, the character's name, your child's username and the dates. You may need them for a report.
3
Report Sexual Content Involving a Child
NCMEC's CyberTipline takes reports of child sexual abuse material and obscene material sent to a child, at report.cybertip.org or 1-800-843-5678 (24 hours). NCMEC says AI-made sexual images of a child should always be reported. Its free Take It Down service helps remove explicit images of a minor, real or AI-made, from participating platforms.
4
Report It in the App, Then Close the Account
Use the app's report tools, then delete the account. Character.AI, for example, says users can delete their account or export their data in the app or on its website.
Don't pass images around. Screenshots of text chats are fine, but never download, forward or post a sexual image of a minor, even to report it. Give NCMEC the web address where it appears instead.
Don't delete first. Save what you need before removing the app or account.
Don't shame your child. These apps are designed to feel like a real relationship, so missing the character is a real feeling.
Don't count on the app to step in. Testing has repeatedly found that crisis responses and parent alerts can fail.
No setting catches every new app. What protects kids most is a parent who knows these apps exist, checks in without judgment, and is the person they come to when a chat gets strange.