ChatGPT for kids: is it safe, and what are the rules by age?

ChatGPT isn't made for children under 13: OpenAI's terms set 13 as the minimum age, require a parent's permission from 13 to 17, and say that in schools, an adult should do the typing for younger children. For teenagers, OpenAI added parental controls and a teen mode with extra protections. Used side by side with a grown-up, with the right settings, ChatGPT can be a good study partner. Left alone with a young child, it's an adult tool.

This post is from Peabody (peabody.kids), a free reading and learning site for kids, free through the end of 2026. Peabody isn't connected to OpenAI and has no open chatbot: its one chat is short and screened, for 3rd grade and up. Everything about ChatGPT below comes from OpenAI's own terms and help pages, opened in October 2026. These settings change often, so check the linked pages before you rely on a detail.

ChatGPT's age rules at a glance

What OpenAI's own rules say

Minimum age. OpenAI's Terms of Use, effective 1 January 2026, say you must be at least 13 (or older, if your country sets a higher age) and that anyone under 18 needs a parent or legal guardian's permission.

Under 13. OpenAI's help article Is ChatGPT safe for all ages? says ChatGPT "is not meant for children under 13". It also warns that ChatGPT may produce answers that aren't right for every age, advises caution "even for those who meet our age requirements", and says that when ChatGPT is used in education with children under 13, an adult must do the actual interacting.

One gap to know about. ChatGPT can be tried at chatgpt.com without an account in supported regions, according to OpenAI's own FAQ. Parental controls work through linked accounts, so they don't cover a child typing into a signed-out browser.

ChatGPT for Teens and age prediction

On 18 August 2026, OpenAI introduced ChatGPT for Teens. If someone says they are 13 to 17, or OpenAI's system estimates they are under 18, their account moves into the teen experience automatically. According to the help page, it adds:

• Extra safeguards that reduce graphic violence, risky viral challenges, sexual, romantic or violent roleplay, and content pushing extreme beauty standards or unhealthy dieting.
• Study mode, which offers hints, steps and follow-up questions instead of just the answer, and reminders that can steer a teen towards it when they seem to be shortcutting homework.
• Reminders to take breaks and that ChatGPT is an AI tool.

Age prediction looks at signals such as the topics someone talks about, the times of day they use ChatGPT and how old the account is. OpenAI says plainly that it can get it wrong. An adult wrongly placed in teen mode can prove their age through Persona, an outside company, with a selfie or ID.

ChatGPT parental controls: what you can and can't do

Parental controls let a parent link their own ChatGPT account to a teen's. Either side can send the invite, by email or phone number, from Settings > Parental controls. One parent can link several teens; each teen links to one parent. According to OpenAI's parental controls page, a linked parent can:

• Keep "Reduce sensitive content" on (on by default).
• Turn off "Improve the model for everyone", which is on by default for teens and lets their conversations be used to train OpenAI's models.
• Turn off saved memories, Voice Mode and image generation.
• Turn on study mode for new chats, or set study hours when new chats start in study mode.
• Set quiet hours, times when the teen can't use ChatGPT at all.
• Get safety notifications in rare, serious cases, such as when trained reviewers see signs of serious self-harm risk.

What a parent can't do: read the teen's conversations. Linking gives no access to chats, history or live activity, and a safety notification says only what kind of concern it is. A teen can also unlink at any time; the parent is told when they do. When the account is identified as 18, the controls end.

That design is a trade-off. It respects a teenager's privacy, and it means the controls are settings, not supervision. Talking with your teen still does more than any toggle.

What to do at each age

Under 6: no chatbot. Common Sense Media's age guidance says no AI chatbots at five and under, because young children can't fully tell AI apart from a real person. If your child is curious, read AI explained for 7-year-olds together, or play Quick, Draw!, where a computer guesses their doodles.

6 to 12: you drive. Common Sense advises limited use beside a parent, and OpenAI's own help page asks for an adult to do the typing for under-13s in education. So use your account, sit together, and let your child tell you what to ask. Read the answers aloud and talk about whether they're right.

13 to 17: their account, linked to yours. Give permission, link parental controls, and look through the settings together rather than in secret. Turn on study hours for homework time. Common Sense suggests supervised use for homework and creativity from 13, never for comfort or feelings. Say that out loud: if something is upsetting them, a person is the right place to go.

Rules for using ChatGPT together

1. Grown-up's account, grown-up's screen, until 13.
2. No personal details. No full name, school, address, photos of faces or friends' names.
3. Check anything that matters somewhere else: a book, a teacher, a trusted site. Chatbots can be confidently wrong.
4. It's a tool, not a friend. It can sound warm. It doesn't know your child.
5. Strange, scary or rude? Stop and tell a grown-up.
6. It asks; you answer. Set it up to quiz, not to do the work (below).

Prompts that make ChatGPT ask questions instead of giving answers

Left to itself, a chatbot explains and a child nods. Flip it, and it becomes a practice partner. Paste one of these at the start of a chat:

• "You're helping a 9-year-old practise. Ask one question at a time about [topic]. Wait for my answer. If I'm wrong, give a small hint, not the answer."
• "Don't solve this for me. Ask me what I already know, then help me find the next step."
• "I just read [book]. Ask me three questions about it, one at a time, and tell me which of my answers you'd push back on."
• "Quiz me on these spelling words. Say each word in a sentence and let me type it."

For teens, study mode does a version of this built in. Use AI as a tutor that asks questions, not one that gives answers has five more prompts and the limits to know.

Alternatives without a chatbot

If the goal is for a child to understand AI, building one teaches more than chatting with one, and it needs no account:

• Teachable Machine (Google, about 8 and up with a grown-up): teach the webcam to tell two things apart. Uses the camera; Google says training can stay on the device.
• AI for Oceans (Code.org, grade 3 and up, sign-in optional): train an AI to sort fish from rubbish, then watch it copy your mistakes.
• Quick, Draw! (Google, 6 and up): every drawing joins a public dataset Google shares with researchers.

The Hour of AI guide turns these into a one-hour plan by age.

And Peabody? Peabody is a reading and learning site, not a ChatGPT replacement. Its Ask box answers from Peabody's own pages and short cards written from open sources, so it may have nothing on your child's question about volcanoes. What it does: AI writes short stories at your child's reading level from ideas they tap or type, and every idea and story is screened before a child sees it. The idea box is one-shot, so there's no back-and-forth with the AI there. The one chat is Talk with Peabody, a short typed chat for 3rd grade and up in a family account. In Peabody Studio a child describes a game and AI builds it, also without a conversation. The math is made and marked by code, not AI. Try it in the library or on the free AI story generator; is AI safe for kids' learning? answers ten safety questions about Peabody and any other app.

In a homeschool

Homeschooling makes the grown-up-types rule easier to keep, because you're already beside your child. For under-13s, try one 15-minute "ask the chatbot" slot a week on your own account: your child picks three questions, you type, and together you check one answer in a book. For a teen, link parental controls and set study hours to match your school mornings. Peabody's part is the reading in between: 10 to 15 minutes a day from the library at your child's level, with no open chatbot. The honest limit: Peabody's Ask box answers only from its own pages and short cards written from open sources, and it doesn't explain a topic the way a tutor can, so it doesn't replace a chatbot, a tutor or you for that. More in AI in homeschooling and on the homeschool hub.

Learn more elsewhere

• OpenAI: managing parental controls: every setting and its default.
• OpenAI: ChatGPT for Teens: what changes for under-18 accounts.
• Common Sense: AI chatbots and your child, by age: the age-and-stage advice quoted above.
• Common Sense AI risk assessments: independent tests of the chatbots and AI products kids use.

Questions parents ask

Is ChatGPT safe for kids?

Not for young children on their own. OpenAI says ChatGPT is not meant for children under 13, may produce answers not right for every age, and in education should be used by an adult for under-13s. For teens 13 to 17, with a parent's permission, its teen mode and parental controls add protections, though no filter catches everything.

What is the minimum age for ChatGPT?

As of October 2026, OpenAI's Terms of Use set the minimum age at 13, or higher where a country requires it. Anyone under 18 needs a parent or legal guardian's permission. Accounts that say or are estimated to be under 18 are placed in ChatGPT for Teens automatically.

Does ChatGPT have parental controls?

Yes, for teens. A parent links their ChatGPT account to their teen's and can manage sensitive-content filters, memory, voice, image generation, model training, study mode and quiet hours, and can receive rare safety notifications. Parents can't read their teen's conversations, and controls don't apply to anyone using ChatGPT signed out.

Can my 10-year-old use ChatGPT?

Not on their own account. OpenAI's minimum age is 13, and its help page says that for under-13s in education an adult should do the interacting. A sensible middle ground is to use it together on your account: your child suggests the questions, you type and read the answers aloud, and you talk about whether they're right.

Can parents see their teen's ChatGPT chats?

No. OpenAI's parental controls don't give access to a teen's conversations, chat history or live activity. In rare cases of serious safety concern, a linked parent may get a notification saying what kind of concern it is, without the conversation itself.

What is a good alternative to ChatGPT for kids?

It depends on the job. To learn how AI works, hands-on tools like Teachable Machine and AI for Oceans need no chat and no account. For reading practice, Peabody writes stories at a child's level with no open chatbot. For homework questions, a grown-up beside the child with ChatGPT set to ask questions is often the best mix.