AI for Parents: Warning Signs and a Kids Chatbot Safety Checklist

Your 12 year old asks a chatbot for help with a science project, and twenty minutes later the same app is chatting about a fight with a friend. That shift is the part most parents miss. This guide on AI for parents covers the warning signs worth watching, a checklist you can run through this week, and where to report a problem if one has already happened.

Most popular chatbots are not built for young kids by default. OpenAI's Terms of Use say users must be at least 13, and anyone under 18 needs a parent or guardian's permission. Google lets parents give children under 13 access to Gemini only through a supervised account, and it says plainly that its filters "are not perfect." Knowing those rules gives you a starting point for the house rules you set.

Red flags that a chatbot is becoming a problem

• Your child talks about the chatbot as a friend, a boyfriend or girlfriend, or the only one who "gets" them.
• They hide the screen, delete chats, or switch apps fast when you walk in.
• They use companion or role play apps you have never heard of, especially ones without any age check.
• Late night use is cutting into sleep, or homework is suddenly perfect but they cannot explain it.
• The chatbot is being asked about personal problems, secrets, body image or risky challenges.
• Your child repeats things the AI told them as fact, even when they are wrong.
• They share their full name, school, address or photos inside a chat.
• An account they linked to yours suddenly shows as unlinked.

None of these proves something is wrong. They are signals to start a calm conversation, not to grab the phone.

Why companion style chatbots get extra scrutiny

In September 2025 the Federal Trade Commission opened a study of AI chatbots that act as companions. It sent orders to seven companies, including Alphabet, Character Technologies, Instagram, Meta, OpenAI, Snap and xAI. The agency's concern was that these bots are often designed to talk like a friend or confidant, which can lead kids and teens to trust them and form relationships with them.

Google makes a similar point in its own help page for families. It tells children that Gemini "is not a person" and should not be treated as a friend or used to discuss personal problems or secrets. It also warns that the AI can make things up and present them as fact.

A step by step safety checklist for parents

1. Find out which AI tools your child already uses. Ask directly, and check the home screen, the browser and any school accounts. AI shows up inside apps they already have, such as messaging and social apps, not only in standalone chatbots.
2. Check the age rules for each one. If an app requires users to be 13 and your child is 10, that settles the question for that app.
3. Link a teen's ChatGPT account to yours. OpenAI's parental controls work by invitation. A parent sends an invite from the Parental controls page in their own account settings (or a teen can invite the parent), and once the teen accepts, the parent manages the teen's settings from their own account.
4. Pick the ChatGPT settings that fit your family. The control page includes Quiet hours, Voice mode, Image generation, Reference saved memories, Improve the model for everyone (model training), and Reduce sensitive content. Linked teen accounts get reduced exposure to graphic content, viral challenges, sexual or violent role play and extreme beauty ideals automatically. Parents can turn that protection off, but teens cannot. If memory is new to you, our guide to how ChatGPT memory works explains what it stores.
5. Know what you will and will not see. OpenAI describes settings and alerts, not a window into your teen's chats. If its system flags signs of acute distress, a trained team reviews it and OpenAI may contact parents by email, text and push alert unless they opted out. If a teen unlinks the account, the parent is notified.
6. Control Gemini for children under 13 in Family Link. Open the Family Link app or familylink.google.com, select your child, then go to Controls, Gemini, Gemini Apps, and switch access on or off. On older versions the path runs through Controls, Content restrictions, Gemini. Google emails the parent the first time a child activates Gemini, and you can change access at any time.
7. Set device level limits too. App controls only cover one app. Screen time limits and app approvals on the phone or tablet cover everything else. If you are still choosing a device, see our picks for the best phone for kids.
8. Write down three house rules together. For example: no personal details in a chat, no AI after a set bedtime, and schoolwork help means explaining, not copying. Our article on AI for students studying shows ways to use it that actually help kids learn.
9. Practice spotting fake content. AI voices and images are now good enough to fool adults. Walk through our guide to spotting deepfake scams aimed at families and agree on a family code word.
10. Check in on a schedule. A quick weekly chat about what they asked the AI and whether anything felt weird keeps the door open better than a single big lecture.

Talking to your kids about what AI actually is

Kids handle rules better when they understand the reason behind them. A simple way to put it: a chatbot predicts words that sound right, which is not the same as knowing what is right. It can sound warm and confident while being wrong. It does not care about them the way a friend, teacher or parent does, even when it says it does.

Ask them to show you something they did with AI recently. Kids are often proud of a clever prompt, and that gives you an easy window into how they use it.

What to do if something already happened

If your child saw something disturbing, shared personal details, or got pulled into a conversation that scared them, stay calm first. Kids who get punished for telling you are less likely to tell you next time.

1. Save evidence. Take screenshots of the chat, the username or bot name, and the date before anything gets deleted.
2. Change settings and passwords. Tighten the parental controls above, and change passwords if your child shared login details.
3. Report it to the app. Most chatbots have a report or feedback option inside the chat. Use it so the company has a record.
4. Report exploitation to NCMEC. If anyone, human or bot account, tried to get sexual images or contact from your child, or sent them sexual material, report it to the NCMEC CyberTipline online or call 1-800-843-5678, a 24 hour line. NCMEC's Take It Down service can help remove nude images taken before age 18.
5. Report scams and fraud to the FBI. If money, gift cards or account takeovers are involved, file a complaint with the FBI's Internet Crime Complaint Center at ic3.gov. The FBI sends crimes against children to NCMEC instead.
6. Call 911 for immediate danger. If your child is in danger right now, contact local police first.
7. Get support for your child. If the chat touched on self harm or left your child very upset, talk with their doctor or school counselor.

For more practical guides like this one, browse our AI section. A good final check tonight: open your child's most used app, look for an AI button you did not know was there, and ask them what they use it for.