Is It Safe for Kids to Use AI Chatbots: A Grandparent Checklist

By Austin Little

Your grandchild asks if they can "use the AI" on your phone to finish a book report. You don't have to panic, and you don't have to say yes on the spot. Here's a calm checklist built from what the companies and the FTC actually say, so you can make a decision you feel good about and that the child's parents will back.

AI disclosure. This page was drafted with AI assistance and edited for Apiary. We don't invent quotes, stats, people, or events.

Not legal advice. This is a practical guide for families. Age rules and parental controls change, so check each service's current terms before relying on anything here. I've linked the official pages I checked on October 4, 2026, and flagged anything I couldn't confirm.

The short answer

Kids can use AI chatbots fairly safely when an adult is involved, the service's age rules are respected, and personal details stay out of the chat. They're riskiest when a child uses one alone, in secret, on an account meant for adults, as a stand-in for a friend or counselor.

That's the whole article in two sentences. The rest is how to get there without lectures, fear, or a fight with the parents.

I'm not going to tell you scary stories or quote alarming percentages. Plenty of what's written about kids and AI is designed to make you anxious enough to click. You don't need that. You need a checklist.

Step zero: the parents' rules come first

This matters more than anything else, so it goes first: you're the grandparent, not the parent. Before you hand over a phone, open an app, or make an account, ask the parents:

• "Are the kids allowed to use AI chatbots? Which ones?"
• "Is there anything you'd like me to avoid?"
• "Do they have their own accounts, or should they only use it with me?"

Some families have already set up controls on the kids' devices. Some have decided "not yet." Some are happy for Grandma to explore it with them. Any of those is a reasonable decision, and respecting it keeps you the trusted adult the child can come to. Undercutting it, even with good intentions, usually backfires.

If the parents say "we haven't thought about it," you can offer to share this checklist. That's a gift, not an overstep.

What the services' own rules say about age

Every major chatbot has a minimum age in its terms. These aren't suggestions. They're the conditions for using the service. Here's what I verified on the official pages today. Rules differ by country and change over time, so treat this as a snapshot.

ChatGPT (OpenAI)

OpenAI's Terms of Use say you must be at least 13 years old (or the minimum age required in your country to consent to use the services), and if you're under 18 you must have your parent or legal guardian's permission.

In plain terms: not for children under 13, and teens need a parent's okay. A grandparent's okay isn't the same thing unless you're the legal guardian.

OpenAI has announced parental-control features for teen accounts. I haven't verified their current details or setup steps, so I'm not describing them here.

Claude (Anthropic)

Anthropic's Consumer Terms of Service say you must be at least 18 years old or the minimum age required to consent in your location, whichever is higher.

In plain terms: Claude's consumer app is for adults. Not for kids or teens on their own accounts.

Gemini (Google)

Google's approach is different. Google's family help page, Guide your child's Gemini Apps experience, says a parent can give a child under 13 (or the applicable age in their country) access to Gemini Apps through a supervised account, managed with the Google Family Link app. Key points from that page:

• A parent must turn access on before a child under 13 can use Gemini Apps with a supervised account, and the parent gets an email when the child first uses it. Parents can turn it on or off anytime in Family Link.
• Google is rolling this out gradually, so access may not be available for every child yet.
• It's not available for supervised accounts in the European Economic Area, Switzerland, or the United Kingdom.
• Google says it adds extra content filters for minors, including filters meant to stop Gemini from claiming to be a person or have human emotions, but also says plainly that its filters "are not perfect" and a child may see content you don't want them to see.
• Google tells parents to remind kids that Gemini isn't a person, shouldn't be treated as a friend or used to discuss personal problems or secrets, and that kids shouldn't share personal information like their address, school name, or family information.

In plain terms: younger children can use Gemini only if a parent sets it up and allows it. That's a parent's decision, made in the parent's Family Link app, not something a grandparent should do on the child's behalf.

Microsoft Copilot

Microsoft's support page Microsoft Copilot for Young People says you must be at least 13 to use Copilot (with a personal Microsoft account), and the minimum may be higher in some countries. The page says that for young people, Copilot doesn't use their conversations for model training, doesn't personalize results, and adds age-appropriate handling for sensitive topics. It also says parents can use tools like Microsoft Family Safety or device settings to block Copilot, set time limits, and choose which features are available.

One wrinkle: that page notes it applies to the older version of the Copilot app, and that a new version came out in August 2026 with its own privacy page.

Other chatbots kids run into

Kids also meet AI inside apps they already use: social media apps, messaging apps, games, and "character" chat apps. I haven't verified the age rules for those today, and they vary a lot. If the child mentions one, look up that app's own terms of service and search for "minimum age."

A quick table (as of October 4, 2026)

What the FTC says about kids' privacy (COPPA)

The Children's Online Privacy Protection Act, usually called COPPA, is a US law enforced by the Federal Trade Commission. The FTC's consumer page Protecting Your Child's Privacy Online explains it for parents. In plain terms:

• It's about kids under 13. The COPPA Rule protects personal information of children under 13 on websites and online services, including apps, that are directed to children. It also applies to general-audience sites that know they're collecting personal information from kids that age.
• Covered services must notify parents and get their approval before collecting, using, or disclosing a child's personal information.
• "Personal information" is broad. The FTC lists things like full name, address, phone number, email, physical location, photos, videos, audio recordings of the child, government identifiers, biometric identifiers, and persistent identifiers like IP addresses that can track a child over time.
• Parents stay in control. Once they consent, parents have the right to review what was collected, withdraw consent, and have the child's information deleted.
• If something looks wrong, the FTC says to report it at ReportFraud.ftc.gov.

Notice two things. First, COPPA gives these rights to parents, which is another reason the parents should be the ones creating accounts and giving consent. Second, COPPA is about data collection. It doesn't make any chatbot "safe" or "approved." A service following COPPA can still give a wrong answer or show content you don't like.

For businesses, the FTC keeps a children's privacy hub with compliance FAQs. You don't need to read it, but it's there if you're curious how the rules apply to companies.

The real risks, calmly stated

Here's what's actually worth managing. None of it is a reason for panic. All of it is a reason for supervision.

1. Confident wrong answers. Chatbots can state false things as fact. Google's own family page says Gemini "can hallucinate and present inaccurate information as factual." For a book report, that means made-up quotes or plot details. For a health question, it could mean bad advice.

2. Content that isn't age-appropriate. Services filter, but even the companies say the filters aren't perfect.

3. Oversharing. Kids happily type their full name, school, address, and family details when they're chatting. Both Google and Microsoft tell young users not to share this kind of information.

4. Treating it like a friend. Chatbots are conversational and can feel personal. Google tells parents to help kids understand Gemini "is not a person" and shouldn't be used to discuss personal problems or secrets. Microsoft says Copilot is "a tool, not a friend." This is the one I'd focus on most, gently. A child who's lonely or upset needs a person.

5. Doing the homework instead of helping with it. Different schools and teachers have different rules about AI. Some allow it for brainstorming, some don't allow it at all. Ask the child what their teacher said, and if they don't know, the answer for now is "help, not do."

6. Accounts that don't fit the child. A 9-year-old signed into Grandpa's adult account isn't getting any of the protections the companies build for minors. It also doesn't follow the terms.

The grandparent checklist

Print this, or keep it on your phone. It's built so you can run through it in a couple of minutes.

Before anyone types anything

• [ ] I asked the parents whether AI chatbots are okay and which ones.
• [ ] I checked the service's minimum age on its official terms or help page, and the child meets it, or the parent set up a supervised account that allows it.
• [ ] The child is using the right account.
• [ ] We're in a shared space. Kitchen table or living room, screen visible. Not behind a closed door.
• [ ] We agreed on the job. "Help me understand fractions," "give me ideas for a story about a dragon," or "quiz me on state capitals." One job.

The privacy rules (say them out loud)

• [ ] No full names, ours or anyone else's.
• [ ] No address, school name, teacher's name, or town.
• [ ] No photos of yourself, family, or friends.
• [ ] No passwords or phone numbers.
• [ ] No secrets. If something's bothering you, tell a person.

These match what Google and Microsoft tell young users on their own help pages, so you're not being overprotective. You're following the manufacturers' instructions.

While we use it

• [ ] We check facts together. "How could we find out if that's true?" is a great habit to build. Look it up in a book, a library website, or an encyclopedia.
• [ ] The child does the thinking. Ask for hints, explanations, and quiz questions, not finished answers.
• [ ] If something weird or upsetting comes up, we stop. Close it, talk about it, and tell the parents. Use the service's report or feedback button if there is one.
• [ ] We keep it short. Twenty minutes, then do something else.

Afterward

• [ ] I told the parents what we did, briefly. "We used it to brainstorm a dragon story, it was fun, it got one fact wrong and we caught it."
• [ ] I signed out if we used a shared device.

A conversation script that doesn't scare anyone

Kids tune out lectures. Try something like this instead, in your own words:

"This thing is really good at sounding smart, but it's a computer program. It doesn't know you, it doesn't have feelings, and it gets stuff wrong sometimes, even when it sounds sure. So we use it like a calculator for words. We don't tell it our name or where we live, and if it ever says something strange, we stop and talk about it. Deal?"

Then make it a game: "Let's see if we can catch it making a mistake." Kids love catching a grown-up thing being wrong, and it teaches the most important skill there is: don't believe something just because it sounds confident.

Good first activities (low risk, high fun)

These work well side by side, and none of them require personal details:

• Story starters. "Give me three ideas for a story about a turtle who wants to fly." Then the child writes the story themselves.
• Explain it like I'm 8. "Explain why the moon changes shape, simply." Then check a library book or a NASA kids' page together.
• Quiz me. "Ask me 10 multiplication questions, one at a time, and tell me if I'm right." The child does the math.
• Silly poems. "Write a funny poem about a dog who loves socks." Then the child writes a better one.
• Plan a project. "What supplies would we need to build a bird feeder from a milk carton?" Then you build it together, off-screen.

Notice the pattern: the AI starts things, and the child and you finish them.

What about local, offline AI?

Some tech-minded families run AI models on their own computer with free apps like Ollama, LM Studio, or Jan, so conversations don't go to a company's servers. That can be good for privacy, but be clear about the trade-off: a local model usually doesn't come with the child-specific filters and age protections that Google and Microsoft describe for minors.

If a family goes this route for kids, the supervision rules above matter even more. For most grandparents, the simpler path is to follow whatever the parents have set up.

What not to worry about

• "Using it once will ruin them." No. Supervised, short, purposeful use is a lot like using a calculator or encyclopedia together.
• "I'm too old to understand this." You understand the most important parts better than most people: don't trust strangers, don't share private details, check before you believe. That's the whole safety model.
• Viral statistics. If a post about kids and AI cites a dramatic number with no source, skip it. The official pages above are better guides than headlines.
• "I need to pay for a kid-safe version." You don't need a paid plan to follow this checklist.

Red flags: when to stop and talk to the parents

Stop and tell the parents, calmly, if you notice:

• The child is using a chatbot in secret or gets upset when asked about it.
• The child describes the chatbot as their best friend or says it "understands them" better than people.
• The chatbot asked for, or the child shared, personal information or photos.
• The child saw something upsetting or inappropriate.
• The child is using an app you've never heard of with AI "characters" or "companions."

None of these mean disaster. They mean it's time for the adults to talk. If a child seems distressed or at risk, put the device down and get the parents, and professional help if needed. A chatbot is never the right counselor for a child in trouble.

If you believe a service collected a child's information or marketed to them in a way that breaks the law, the FTC's consumer page says to report it at ReportFraud.ftc.gov.

Quick answers

"Is ChatGPT safe for kids?" OpenAI's terms require users to be at least 13, with a parent's or guardian's permission under 18. For younger kids, no. For teens, it depends on the parents' permission and supervision.

"What age can kids use AI chatbots?" It depends on the service. ChatGPT and Microsoft Copilot list 13+. Claude lists 18+. Gemini lets parents enable access for children under 13 through Family Link where it's available. Always check the current official terms.

"Can I let my grandchild use AI on my phone?" Ask the parents first, check the service's age rules, and if you do it, sit together and keep personal details out.

"Does COPPA make chatbots safe for kids?" No. COPPA, enforced by the FTC, is about collecting personal information from children under 13 and requires parental consent for covered services. It's a privacy law, not a safety guarantee.

"What's the biggest risk?" Probably a child treating a chatbot as a confidant. Keep reminding them it's a tool, not a person, and that worries go to people.

Bottom line

AI chatbots aren't something to fear, and they're not something to hand a child and walk away from. Respect the parents' rules, check the age on the official terms, keep personal details out, sit together, and check facts as a team. Do that and the chatbot becomes one more kitchen-table tool, like the dictionary or the calculator, with you right there.

You already have the most important safety skill: you know not to trust something just because it sounds sure of itself. Pass that on.

References

• OpenAI — Terms of Use: https://openai.com/policies/terms-of-use/ (fetched 2026-10-04)
• Anthropic — Consumer Terms of Service: https://www.anthropic.com/legal/consumer-terms (fetched 2026-10-04)
• Google For Families Help — Guide your child's Gemini Apps experience: https://support.google.com/families/answer/16109150?hl=en (fetched 2026-10-04)
• Microsoft Support — Microsoft Copilot for Young People: https://support.microsoft.com/en-us/microsoft-copilot/microsoft-copilot-young-people (fetched 2026-10-04)
• FTC Consumer Advice — Protecting Your Child's Privacy Online: https://consumer.ftc.gov/articles/protecting-your-childs-privacy-online (fetched 2026-10-04)
• FTC — Children's Privacy business guidance: https://www.ftc.gov/business-guidance/privacy-security/childrens-privacy (fetched 2026-10-04)
• FTC — ReportFraud: https://reportfraud.ftc.gov/