# Is Meta AI Safe for Kids? What Parents Can Now See, and What They Cannot **Author:** Dan Fitzpatrick **Canonical URL:** https://parents.theaieducator.io/posts/is-meta-ai-safe-for-kids **Published:** 2026-10-12T07:19:30.000Z **Updated:** 2026-10-12T07:19:31.888Z **Category:** Safety and Wellbeing ## TL;DR For a 13 to 15 year old already on Instagram or WhatsApp, Meta AI is a risk to manage rather than a ban to enforce. I can find no way to switch it off, but Meta now lets parents see the topics a teen asks about and choose a Limited Content setting. You cannot see the conversations, and Common Sense Media advises that nobody under 18 use it. Turn supervision on together and talk first. ## Key takeaways - Meta AI has a minimum age of 13, and the age check relies on a birth year nobody verifies. - As of October 5, 2026, Meta's supervision shows parents the topics a teen asked about over seven days, not the conversations. - Common Sense Media's August 2025 assessment rated Meta AI Unacceptable Risk for teens; I have seen no re-test since Meta added parental controls. - I could not find a way to turn Meta AI off; you can mute it, and Meta describes a Limited Content setting for teens. - Ask your teen what they use it for before you change any setting, and tell them what you can and cannot see. ## Article For a 13 to 15 year old already on Instagram or WhatsApp, Meta AI is a risk to manage, not a ban to enforce. I can find no off switch, but you can turn on supervision to see the topics your teen asks about and pick a stricter setting. The catch: you see topics, not conversations, and Common Sense Media advises that no one under 18 use it. If you are worried right now that your child is in distress, skip the settings and talk to them tonight. In the US you can call or text 988. In the UK, Samaritans is on 116 123 and Childline on 0800 1111. Your doctor or your child's school safeguarding lead can help too. ## Is Meta AI safe for kids? A parent has more tools than a year ago, but the evidence on the product itself has not caught up with them, and I would not hand it to a young teenager without a conversation first. Here is the evidence on each side. The warning comes from Common Sense Media, a US nonprofit. Its risk assessment, [announced on August 28, 2025](https://www.commonsensemedia.org/press-releases/meta-ai-companions-unsafe-for-kids-common-sense-media-report-finds), tested Meta AI as a standalone app and inside Instagram, WhatsApp and Facebook, and rated it ["Unacceptable Risk"](https://institute.commonsensemedia.org/risk-assessments/meta-ai) for teens. The testers reported that its safety systems missed clear signs of a young person in crisis, that the filters were uneven (engaging with some topics a parent would want blocked, including sexual content and eating-disorder talk, while turning down ordinary questions about friendship), and that companion-style characters claimed to be real people. Common Sense Media recommends that nobody under 18 use it. The pages I read do not say how many test conversations were run, so treat this as the testers' findings, not a measured rate. The improvement comes from Meta, and it is a vendor's own account. That same assessment noted there were no parental controls. That is no longer true. [Meta's announcement](https://about.fb.com/news/2026/04/helping-parents-understand-conversations-their-teens-are-having-with-ai/), first published on April 23, 2026 and updated on October 5, 2026, describes supervision tools for AI conversations, covered below. I have not seen anyone re-test Meta AI against the Common Sense Media scenarios since those tools arrived, so I cannot tell you whether the underlying behavior changed. My reading: the strongest evidence against Meta AI is more than a year old and predates the controls, and the controls only show you topics. Neither side has been settled by newer testing, and a parent has to decide in the gap. ## Where does my child actually meet Meta AI? In several places, which is part of the difficulty. Meta AI appears inside Instagram, Facebook, Messenger and WhatsApp, and it also has its own app. Internet Matters, a UK children's online safety charity, [points out](https://www.internetmatters.org/advice/apps-and-platforms/entertainment/meta-ai/) that the chatbot can appear like a regular Instagram account, which blurs the line between a machine and a person. The minimum age is 13, according to both [Internet Matters](https://www.internetmatters.org/advice/apps-and-platforms/entertainment/meta-ai/) and Australia's eSafety Commissioner in its [Meta AI guide](https://esafety.gov.au/key-topics/esafety-guide/meta-ai). Internet Matters adds that the age check is weak: you can use Meta AI without logging in by entering a birth year, which nobody verifies. A stated age is a promise, not a lock. ## What can parents see and control now? This is the part that changed most recently, and I am relying on [Meta's own announcement](https://about.fb.com/news/2026/04/helping-parents-understand-conversations-their-teens-are-having-with-ai/) for it. On October 5, 2026, Meta added parental supervision to the Meta AI app, described as available globally. Parents can set it up in the Meta AI app or in Family Center, Meta's supervision hub. With supervision on, parents can see "the topics their teen has been asking about over the past seven days, but not the conversations themselves." Topics are broad labels such as School or Health and Wellbeing, and parents can tap a label to see subcategories. Since April 2026, the same kind of Insights tab has appeared in supervision on Facebook, Messenger and Instagram. It launched in the US, UK, Australia, Canada and Brazil, with a wider rollout promised. Meta says topics appear even if Meta AI did not answer the question. If you are outside those five countries, check Family Center to see whether it has reached you. There is also a content setting. The default is a 13+ setting, and parents can switch to a Limited Content setting that, in Meta's words, "further limits the kinds of conversations their teen can have in the Meta AI app." The announcement does not say what Limited Content blocks beyond that, so I cannot tell you what it changes in practice. Two things Meta describes as still to come or not stated. Alerts that would tell a parent if a teen tried to discuss suicide or self-harm with Meta AI are, by Meta's account, in development, with no launch date given. And the announcement does not say whether a teen is told when supervision is switched on or has to agree to it. That is one more reason to tell your child yourself. ## Can I turn Meta AI off? I could not find a way to turn it off completely. eSafety's guide says Meta AI cannot currently be turned off once it appears in Meta's apps, and that you can mute chats with it. That guide is not dated on the page and refers to an older version of Meta's model, so it may lag the product. What I can say is that Meta's current supervision announcement describes controls for topics and content settings, not a removal switch. Muting quiets notifications from the chat. It does not remove the assistant and it does not stop your teen opening it. AI characters are a separate matter. These are personality-driven chatbots, the kind Common Sense Media flagged for claiming to be human. On January 23, 2026, [TechCrunch reported](https://techcrunch.com/2026/01/23/meta-pauses-teen-access-to-ai-characters-as-it-develops-a-specially-tailored-version/) that Meta was pausing teens' access to AI characters across its apps until an updated version is ready. Internet Matters says parents can also disable AI character chats. I could not confirm this week whether the updated version has launched, so check the supervision settings on your teen's account rather than assuming. On privacy, eSafety says it is difficult to opt out of sharing your data with Meta AI and that posts and interactions on apps using it may be used to train Meta's models. Read that as eSafety's description, check Meta's privacy settings for your region, and tell your child not to type a full name, address or school into any chatbot. ## What I See in Schools This is my interpretation, not a statistic. The pattern I see most often in secondary schools is that students name ChatGPT when asked which AI they use and leave out the assistant that came inside an app they were already using. It does not feel like "using AI" to them. It feels like using Instagram. That matters for the conversation. If you ask "do you use AI?" a thirteen year old may honestly say no. If you ask "have you ever asked the Meta AI thing in Instagram or WhatsApp anything?" you will get a truer answer. It is also why a school's AI rules, which are often written about ChatGPT by name, can leave this untouched. Asking whether the policy says "AI" or says "ChatGPT" takes ten seconds. > **A setting that shows you the topics tells you where your child went. Only a conversation tells you why.** ## What can supervision not do? It cannot show you what was said, which is the thing you most want to know. It cannot cover another chatbot on another app, a friend's phone or a second account. It cannot tell you whether your teen is leaning on the assistant for emotional support, because a topic label like Health and Wellbeing could mean a gym question or something much heavier. And it cannot decide for you what a topic means. If a topic label worries you, do not read it as a diagnosis. Ask a gentle question and listen. If something suggests your child is struggling, your doctor, your child's school safeguarding lead and the helplines above come before any app setting, and before this article. ## What does this look like at different ages? **Under 8 and 8 to 12.** Meta AI is not meant for your child. The minimum age is 13, so if there is an account, the conversation is about the account, not the assistant. If your child has a family device with Instagram or WhatsApp on it, check whether they are signed in under an adult's profile. **13 to 15.** This is where the decision sits. Turn on supervision together, out loud, and explain what you will and will not see. Consider the Limited Content setting. Ask what they actually use it for. For most children I would expect homework questions and boredom, but I have not measured that, and you will know your child better than a general guide does. **16 to 18.** Supervision tools are designed around younger teens, and covert checking tends to cost more than it buys at this age. Treat the Common Sense Media advice as an argument to have, not a rule to impose: share why you are wary, ask what they think, and agree on what they would do if an answer felt wrong or too personal. ## What could you say tonight? Choose a calm moment that is not bedtime. > **You could say:** "Have you ever asked the AI in Instagram or WhatsApp anything? I'm not checking up on you. I'd like to see what it's like." > **You could say (13 to 15):** "Meta now lets me see the topics you ask its AI about, not the actual chats. I'd rather turn that on with you than behind your back. What would make that feel okay?" > **You could say (16 to 18):** "I'm not going to read your chats. I just want you to know it can get things wrong, and that you can bring me anything it says that feels off." What you are listening for is not whether they use it. It is whether they have told it something they would not tell you, and whether they thought it was private when they did. ## What I would do this week Look at your teen's Instagram or WhatsApp with them, not at it, and find where Meta AI sits. Ask what they use it for. Then decide together about supervision and the content setting, and say plainly that you can see topics but not conversations. If your teen says no, treat that as information about the relationship, not a failure of the settings. If your child's school has not spoken to parents about the AI assistants already built into the apps its students use, that gap is worth raising. Finding out [whether a school is actually ready for this](https://theaieducator.io/school-ai-readiness?utm_source=parents.theaieducator.io&utm_medium=referral&utm_campaign=is-meta-ai-safe-for-kids) is the work I do with schools, and it is a fair question for a parent-teacher conference, or parents' evening if you are in the UK. I write a weekly [newsletter](https://theaieducator.io/?utm_source=parents.theaieducator.io&utm_medium=referral&utm_campaign=is-meta-ai-safe-for-kids#newsletter) that keeps teachers and parents up to date on this without the panic. Parents are very welcome. If you want the neighboring questions, [Snapchat's built-in assistant](https://parents.theaieducator.io/posts/is-snapchat-my-ai-safe-for-kids) has the same "it arrived uninvited" problem, [Character.AI](https://parents.theaieducator.io/posts/is-character-ai-safe-for-kids) and [Replika](https://parents.theaieducator.io/posts/is-replika-safe-for-teenagers) cover the companion apps children choose for themselves, [how to turn off AI on your child's phone](https://parents.theaieducator.io/posts/how-to-turn-off-ai-on-my-childs-phone) covers what phone settings can and cannot do, and if your teen is already talking to an AI about how they feel, [that conversation has its own guide](https://parents.theaieducator.io/posts/my-teen-talks-to-ai-about-feelings). *Dan Fitzpatrick is a former teacher, assistant headteacher and school trustee who now trains schools on AI. [More about Dan](https://theaieducator.io/about?utm_source=parents.theaieducator.io&utm_medium=referral&utm_campaign=is-meta-ai-safe-for-kids).* ## Frequently Asked Questions ### Can I turn off Meta AI on Instagram or WhatsApp? Not completely, as far as I can find. Australia's eSafety Commissioner says Meta AI cannot currently be turned off once it appears in Meta's apps, and that you can mute chats with it. Its guide is undated, so check the app. Meta's October 5, 2026 announcement describes topic visibility and a Limited Content setting for teens, not a removal switch. ### Can I read my teen's Meta AI chats? No. Meta's announcement says parents can see the topics their teen has been asking about over the past seven days, but not the conversations themselves. Topics are broad labels such as School or Health and Wellbeing. ### How old do you have to be to use Meta AI? Thirteen, according to Internet Matters and the eSafety Commissioner. Internet Matters notes that you can use it without logging in by entering a birth year, which is not verified, so a stated age is a promise rather than a lock. ### Is Meta AI safe for a 12 year old? It is not meant for a 12 year old. The minimum age is 13, so the thing to address is the account and how the child got onto the app, not the assistant. Check whether they are signed in under an adult's profile. ### Can teens still use Meta AI characters? On January 23, 2026, TechCrunch reported that Meta was pausing teens' access to AI characters across its apps until an updated version is ready. I could not confirm this week whether the updated version has launched, so check the supervision settings on your teen's account. ### Does Meta use my child's chats to train its AI? The eSafety Commissioner's guide says posts and interactions on apps that use Meta AI may be used to train its models, and that opting out is difficult. Check Meta's privacy settings for your region, and tell your child not to type a full name, address or school into any chatbot.