AI Chatbots and Kids: Ages, Parental Controls and What to Say

Each chatbot sets its own minimum age, the parental controls around them are settings rather than locks, and a calm conversation with a child does more than either. This page is for parents and anyone teaching a child at home: what the main services say in their own words, what their controls do, four things worth talking about, and what the law says.

What each service says about age

The rules as each company writes them. Terms change, so each citation carries the date it was read.

• ChatGPT (OpenAI). The terms say: “You must be at least 13 years old or the minimum age required in your country to consent to use the Services. If you are under 18 you must have your parent or legal guardian’s permission to use the Services.”OpenAI, Terms of Use, effective 1 January 2026, the version for users outside the EEA, Switzerland and the UK, read at source 18 Sep 2026. Accounts that OpenAI believes belong to someone under 18 are moved into a teen version: “If your account is eligible and account-provided age information, verified age, or age prediction indicates you are under 18, ChatGPT automatically enables ChatGPT for Teens.”OpenAI Help Center, ChatGPT for Teens, read at source 18 Sep 2026 in a browser, because the page refuses automated fetches. The page says the teen version is still rolling out. In Australia, OpenAI says it began making the teen version the default for users it identifies as 13 to 17 in August 2026.OpenAI, Introducing the Australian Youth Safety Blueprint, 18 Sep 2026, read at source 22 Sep 2026 in a browser: “In August, we began rolling out ChatGPT for Teens in Australia, a new default experience for users identified as aged 13 to 17”.
• Gemini (Google). For a child’s own account: “You must be 13 (or the applicable age in your country) or over to use Gemini with a personal or school Google Account”. Younger children can use it only through a supervised account: “A parent must enable access before their child under 13 (or the applicable age in your country) can use Gemini Apps with a supervised account.”Google, What you need to sign in to Gemini Apps and Guide your child’s Gemini Apps experience, both read at source 18 Sep 2026. The second page also says supervised-account access is not available in the European Economic Area, Switzerland and the UK.
• Meta AI. The terms say “you cannot access AIs if you are under the age of 13 (or such greater age required in your country or territory)”.Meta, Meta AIs Terms of Service, effective 13 May 2026, read at source 18 Sep 2026 in a browser without logging in. The UK, Brazil and the European Region have separate versions.
• Character.AI. The terms say: “If you are under 13 years old OR if you are under 16 years old and a citizen or resident in the European Economic Area (EEA) or the United Kingdom (UK), do not sign up for the Services – you are not authorized to use them.”Character.AI, Terms of Service, last updated 27 August 2025, read at source 18 Sep 2026 in a browser. Separately, the company announced in October 2025 that it would be “removing the ability for users under 18 to engage in open-ended chat with AI on our platform.”Character.AI, Taking Bold Steps to Keep Teen Users Safe on Character.AI, 29 Oct 2025, read at source 18 Sep 2026. A follow-up post of 21 Nov 2025 says removal would begin with under-18 users “in the US, with other markets following in short order.”
• Claude (Anthropic). The terms say: “You must be at least 18 years old or the minimum age required to consent to use the Services in your location, whichever is higher.”Anthropic, Consumer Terms of Service, effective 8 October 2025, read at source 18 Sep 2026. Its help centre adds: “We have safety systems in place to detect if people under 18 may be using Claude and we’ll disable accounts based on indicators of minor activity.”Anthropic, Age assurance on Claude, Claude Help Center, read at source 18 Sep 2026. An account flagged this way can be reinstated by passing an age check with Yoti, an outside provider, and Anthropic says it receives only a pass or fail.Anthropic, Age assurance on Claude, read at source 22 Sep 2026: “You’ll have the option to verify your age using Yoti, a third-party age verification platform.” and “Anthropic never sees your ID or image; we receive only a pass/fail result”.

These are conditions in a contract, backed more and more by systems that estimate age. A birthday typed at sign-up is only as true as the person typing it, and an estimate can be wrong either way.Reasoning, September 2026 — drawn from the companies’ own wording above: OpenAI and Anthropic both describe inferring age from signals, not only asking for it.

What parental controls do, and what they do not

ChatGPT lets a parent and a teen link accounts, switch features off and set quiet hours, “times when your teen cannot use ChatGPT”. What linking does not do: “Linking accounts does not give a parent or guardian access to a teen’s conversations, chat history, or real-time activity.” And on its alerts: “Safety notifications are not real-time monitoring. They may not identify every concern and do not replace professional care or emergency services.”OpenAI Help Center, Managing parental controls in ChatGPT, read at source 18 Sep 2026.

One setting is worth checking. “Improve the model for everyone” controls whether a teen’s conversations can be used to improve ChatGPT, and the page lists it as on by default.OpenAI Help Center, Managing parental controls in ChatGPT, read at source 18 Sep 2026: “Controls whether your teen’s conversations can be used to help improve ChatGPT models.”, listed with “Default: Enabled”. See what happens to what you type.

Google’s Family Link turns Gemini on or off for a younger child, and Google says of its extra filters for minors: “These filters have been designed for minors and try to limit access to inappropriate content, but they’re not perfect.”Google, Guide your child’s Gemini Apps experience, read at source 18 Sep 2026. Meta shows supervising parents the topics, not the messages, in a new Insights tab: “parents will be able to see the topics their teen has been asking Meta AI about in that specific app over the past week.”Meta, Helping Parents Understand the Conversations Their Teens Are Having With AI, 23 Apr 2026, read at source 18 Sep 2026. In January 2026 Meta also said it was “temporarily pausing teens’ access to existing AI characters globally” while it rebuilt them.Meta, Empowering Parents, Protecting Teens: Meta’s Approach to AI Safety, update dated 23 Jan 2026, read at source 18 Sep 2026.

Four things to talk about

It can be wrong, confidently. Google tells parents that Gemini “can hallucinate and present inaccurate information as factual.”Google, Guide your child’s Gemini Apps experience, read at source 18 Sep 2026. A habit for any age: ask where an answer came from, then check it. See how to check an AI citation.

It is not a friend, and it does not keep secrets the way a friend does. The same Google page asks parents to remind children that “Gemini should not be treated as a friend or used to discuss personal problems or secrets.” The US Federal Trade Commission put the risk this way: chatbots “generally are designed to communicate like a friend or confidant, which may prompt some users, especially children and teens, to trust and form relationships with chatbots.”Google, same page, read at source 18 Sep 2026; Federal Trade Commission, FTC Launches Inquiry into AI Chatbots Acting as Companions, 11 Sep 2025, read at source 18 Sep 2026.

Keep personal details out. Google’s own advice is to tell a child “not to share sensitive personal information like their address, school name, or family information in their chats with Gemini Apps.”Google, Guide your child’s Gemini Apps experience, read at source 18 Sep 2026. The same goes for photos that show a name, and for other people’s details.

Companion apps are different. An assistant answers questions; a companion app is built to be a relationship. When Character.AI ended open-ended chat for under-18s, it gave as a reason questions about “how open-ended AI chat in general might affect teens, even when content controls work perfectly.”Character.AI, Taking Bold Steps to Keep Teen Users Safe on Character.AI, 29 Oct 2025, read at source 18 Sep 2026. The evidence is on AI companions, and its question applies here too: is it a place to rest before going back to people, or has it become the place?

If a child tells you something worrying, OpenAI’s advice to parents is sound whichever app it was: “Check in with your teen. Ask about their immediate safety, listen without judgment, and help them contact someone they trust, a qualified professional, or a crisis helpline.”OpenAI Help Center, Managing parental controls in ChatGPT, read at source 18 Sep 2026.

If you are teaching at home

For younger children, OpenAI is explicit: “if you are using ChatGPT in the education context for children under 13, the actual interaction with ChatGPT must be conducted by an adult.”OpenAI Help Center, Is ChatGPT safe for all ages?, read at source 18 Sep 2026 in a browser, because the page refuses automated fetches. In practice: sit side by side, type the questions yourself, and ask the child to judge the answers. For older students, AI in schools explains why detector scores are not proof and drafts tell you more.

What the law says, where this site could read it

US children’s privacy (COPPA). The rule is the Federal Trade Commission’s. It covers services “directed to children under 13 years of age” and those with “actual knowledge that they are collecting personal information online from a child under 13 years of age.” The FTC’s guidance adds: “As a general rule, operators must get verifiable parental consent before collecting personal information online from children under 13.”Federal Trade Commission, Children’s Online Privacy Protection Rule (“COPPA”) and Complying with COPPA: Frequently Asked Questions, both read at source 18 Sep 2026. That is the US background to the age of 13 in the terms above, though none of them says why it chose that age.Reasoning, September 2026 — the link between COPPA and the terms is this page’s inference, not something any company states. The rule was amended in 2025, and covered companies had until 22 April 2026 to comply fully with most of the changes. In February 2026 the FTC added that it would not bring COPPA cases against general- and mixed-audience services that collect personal information only to work out a user’s age, if they meet its conditions.Federal Trade Commission, FTC Finalizes Changes to Children’s Privacy Rule, 16 Jan 2025, read at source 22 Sep 2026: “Entities subject to the final rule will have one year from that publication date to come into full compliance with amendments that do not specify earlier compliance dates.” The FTC’s Federal Register notice page dates the publication “April 22, 2025”; one year on is 22 April 2026. FTC Issues COPPA Policy Statement to Incentivize the Use of Age Verification Technologies, 25 Feb 2026, read at source 22 Sep 2026: the Commission “will not bring an enforcement action under the Children’s Online Privacy Protection Rule (COPPA Rule) against certain website and online service operators that collect, use, and disclose personal information for the sole purpose of determining a user’s age via age verification technologies.”

California’s companion chatbot law (SB 243). Approved on 13 October 2025, it requires operators, for a user they know is a minor, to disclose that the user is talking to AI and to “Provide by default a clear and conspicuous notification to the user at least every three hours for continuing companion chatbot interactions that reminds the user to take a break and that the companion chatbot is artificially generated and not human.” It also requires operators to disclose “that companion chatbots may not be suitable for some minors.”California Legislature, SB-243 Companion chatbots, chaptered text, read at source 18 Sep 2026: “Approved by Governor October 13, 2025.” The law defines companion chatbots narrowly and excludes, for example, customer-service bots.

California’s newer laws (SB 1119 and SB 867). Signed on 10 September 2026. SB 1119, which the Governor’s office calls “Adam’s Law”, requires companion-chatbot operators to carry out a child-safety risk assessment before offering a new or substantially modified chatbot from 1 July 2027, and to have an independent child-safety audit by 1 January 2029 or before launch, whichever is later. SB 867 bans the sale of toys that include a companion chatbot until 1 January 2031.California Legislature, SB-1119 Companion chatbots: children’s safety, chaptered text, read at source 22 Sep 2026: “Approved by Governor September 10, 2026”; operators must, “beginning July 1, 2027, before making a new or substantially modified companion chatbot available to users in the state”, carry out a risk assessment; and “On or before January 1, 2029, or before an operator first makes a companion chatbot publicly available, whichever is later, an operator shall ensure the performance of a child safety audit.” SB-867 Toys: companion chatbots, chaptered text, read at source 22 Sep 2026: “No person shall manufacture, sell, or exchange, possess with intent to sell or exchange, or expose or offer for sale or exchange to any retailer any toy that includes a companion chatbot.” and “This section shall remain in effect only until January 1, 2031, and as of that date is repealed.” The name comes from the Office of the Governor of California, 10 Sep 2026, read at source 22 Sep 2026: “Named after Adam Raine”. SB 1119 also removes the SB 243 duties above that apply when an operator knows a user is a minor, including the break reminder, in favour of its own child-safety rules.California Legislature, SB-1119, Legislative Counsel’s digest, read at source 22 Sep 2026: “This bill would delete those provisions imposing requirements on an operator if the operator knows the user is a minor.” The amended section 22602 in the bill keeps only the not-human notice and the suicide and self-harm protocol.

A federal inquiry. In September 2025 the FTC ordered seven companies, including Alphabet, Character Technologies, Meta, OpenAI and Snap, to explain how they measure, test and monitor possible harms from their chatbots to children and teens. It describes this as a study under authority to “conduct wide-ranging studies that do not have a specific law enforcement purpose.”Federal Trade Commission, FTC Launches Inquiry into AI Chatbots Acting as Companions, 11 Sep 2025, read at source 18 Sep 2026. It is a request for information, not a finding, and no findings have been published yet: the FTC’s page for the inquiry lists only a resolution, a model order and a cover letter.Federal Trade Commission, 6(b) Orders to File Special Report … AI Companion Products or Services, read at source 22 Sep 2026. The page lists three documents, all from September 2025: the “Resolution Directing Use of Compulsory Process”, a “Model Order to File Special Report” and an “AI Companion Chatbot Cover Letter”.

A short plan

1. List the apps, including AI features inside apps your child already uses.
2. Read each age rule on the company’s own page.
3. Set up controls together, so the child knows what is on and why.
4. Have the four conversations, more than once.
5. Look again every few months. Several services on this page changed their teen rules in the past year.

The habit underneath is the one on the family guide: agreed rules the whole household follows, in place of a judgement made in the moment.Reasoning, September 2026 — summarises this page’s sections above; the claim about changes rests on the dated announcements and terms cited there.

What this page could not verify

• How well any age check works. No study of how often children get past them was read.
• How many children use chatbots. No usage figure is quoted, because none was read at source.
• Whether announced changes are complete everywhere. Character.AI said other markets would follow the US; Meta called its pause on teen AI characters temporary. No later statement on either was found.
• Law outside the US. Several terms raise the age in some countries; local law there was not checked.
• OpenAI’s main terms page (/policies/terms-of-use/) refused automated access; the version quoted is the one OpenAI publishes for users outside the EEA, Switzerland and the UK.