Set Up Parental Controls for the AI Chatbots Your Child Uses

The tip

If your child or teenager uses an AI chatbot, spend twenty minutes this week finding out which one, then switch on the family controls that the company behind it offers. The three most common general chatbots handle young users in three very different ways:

• ChatGPT lets a parent link their own account to their teen's account and manage the teen's settings from a page called Parental controls in account settings.
• Google Gemini can be turned on or off for a child under 13 (or the applicable age in your country) through the Google Family Link app, and Google adds extra content filters for signed in minors.
• Claude is not meant for children or teens at all. Anthropic says you must be at least 18 to create and use a Claude account.

This guide walks through each one: where the controls are, what each setting actually does, what it does not do, and the mistakes that leave gaps. Everything here comes from the companies' own help pages and announcements, listed in the sources at the end. Menus change, so if a label on your screen looks slightly different, look for the nearest match rather than giving up.

Why it matters

AI chatbots are not like a search engine. They talk back, remember things and can feel like a person. Google says this plainly in its guidance for parents: Gemini "is conversational and can feel like a human," and parents should help children understand that "it's an AI tool, not a person to confide in." Google also warns that Gemini "can hallucinate and present inaccurate information as factual."

The companies also admit that their filters have limits. Google writes that its filters "try to limit inappropriate content from appearing, but they're not perfect," and that "your child may encounter content you don't want them to see." OpenAI says that guardrails "help, but they're not foolproof and can be bypassed if someone is intentionally trying to get around them."

That is the honest starting point. Parental controls do not make a chatbot safe on their own. What they do is reduce some risks, remove some features you may not want your child to have yet, and in some cases tell you when something serious seems to be happening. Combined with a conversation at home, that is worth twenty minutes.

Step 1: Find out which chatbots your child actually uses

Before changing any setting, make a short list. Ask your child directly, and check the devices they use.

1. Ask. A relaxed question works better than an interrogation: "Which AI apps do you use for homework or for fun?"
2. Check the phone or tablet home screen and app library for ChatGPT, Gemini, Claude and other AI apps.
3. Check the browser. Chatbots also run in a web browser, so an app list alone can miss them.
4. Check school accounts. Google says Gemini can be used with a school Google Account only if the school's administrator has enabled it. If your child uses AI through a school account, the school controls that access, not you.
5. Check smart speakers and displays in your home. Google's Gemini for Home voice assistant runs on compatible devices such as Google Nest Audio and Google Nest Hub Max in select countries and languages.

Write down each chatbot and which account your child uses with it. You now know which of the sections below apply to your family.

ChatGPT: link your account to your teen's account

OpenAI announced parental controls for ChatGPT on September 29, 2025, and said they were "available to all ChatGPT users." The system works by linking two accounts: yours and your teen's. You need your own ChatGPT account to do this.

How to link accounts

OpenAI describes the process like this:

1. A parent or guardian sends an invite to their teen to connect accounts. OpenAI's announcement shows this starting from a Parental controls section in the account settings menu, with a button to add a family member.
2. The teen accepts the invite. Until they do, the accounts are not linked and none of the controls apply.
3. Once linked, you manage the teen's settings from your own account, on what OpenAI calls "a simple control page in account settings."

It also works in the other direction. OpenAI says "teens can also invite a parent to connect." If your teen is happy to set this up, letting them send the invite can make it feel less like surveillance and more like a shared decision.

What each ChatGPT control does

After linking, OpenAI lists these options for parents. They are optional, and OpenAI describes them as "flexible so families can choose what works best for them."

• Quiet hours. Set specific times when ChatGPT can't be used. This is useful for bedtime or homework periods where you want a break from the app. It controls ChatGPT only, not the phone as a whole.
• Voice mode. Turning it off removes the option to talk to ChatGPT by voice. Your teen can still type.
• Memory. Turning it off means ChatGPT won't save and use memories when responding. In OpenAI's screenshots this toggle is labelled Reference saved memories. With it off, ChatGPT does not build up a running picture of your teen across chats.
• Image generation. Turning it off means ChatGPT won't have the ability to create or edit images for your teen.
• Model training. You can opt out so your teen's conversations won't be used to improve the models powering ChatGPT. In OpenAI's screenshots this toggle is labelled Improve the model for everyone.

Extra content protections that switch on when you link

Linking does more than unlock the settings above. OpenAI says a linked teen account automatically gets additional content protections, "including reduced graphic content, viral challenges, sexual, romantic or violent roleplay, and extreme beauty ideals." In OpenAI's screenshots this appears as Reduce sensitive content, switched on.

Two details matter here:

• Your teen cannot change this setting. OpenAI says "teen users cannot make changes."
• You can. OpenAI says parents have the option to turn it off. Unless you have a clear reason, leave it on.

Safety notifications

OpenAI says it has added protections that help ChatGPT recognise potential signs that a teen might be thinking about harming themselves. When its systems detect potential harm, "a small team of specially trained people reviews the situation." If there are signs of acute distress, OpenAI says it will contact parents "by email, text message and push alert on their phone, unless they have opted out."

OpenAI also says it shares only the information needed for parents or emergency responders to protect a teen's safety, and admits that it "might sometimes raise an alarm when there isn't real danger." Make sure the email address and phone number on your own ChatGPT account are ones you actually check.

What ChatGPT parental controls do not do

• They do not apply until your teen accepts the link. An invite that sits unanswered changes nothing.
• They do not cover other accounts. The controls apply to the linked account. If your teen also uses a different ChatGPT account, the controls do not follow them there.
• They are not described as a way to read chats. OpenAI's announcement lists settings and safety notifications. It does not describe parents reading their teen's conversations, so do not assume you can.
• They can be undone by your teen. OpenAI says that "if a teen unlinks their account, their parent will be notified." That notification is your cue to talk, not to panic.

Google Gemini: manage access for a child under 13 with Family Link

Google handles children differently. For a child under 13 (or the applicable age in your country) with a Google Account supervised through Google Family Link, the parent decides whether Gemini Apps can be used at all.

Check availability first

Google lists some limits on its help page:

• Google says it is "gradually releasing access to Gemini Apps for supervised accounts," so your child may not have access yet even if you switch it on.
• Google says Gemini Apps "are not available for supervised accounts in the European Economic Area, Switzerland, and the United Kingdom."
• Google's sign in help page also states that a Google Account managed by Family Link still can't access the Gemini web app. If your child can use Gemini on a phone but not in a browser, this is the likely reason.

How to turn Gemini on or off for your child

Google says a parent must enable access before a child under 13 (or the applicable age in your country) can use Gemini Apps with a supervised account, and that you'll receive an email notification when your child first activates the app. You can change access at any time:

1. Go to familylink.google.com or open the Family Link app.
2. Select your child.
3. Tap Controls, then Gemini, then Gemini Apps.
4. Turn Gemini Apps on or off.

On older versions of Family Link, the path is Controls, then Content restrictions, then Gemini, then Gemini Apps.

If your child sees a message saying Gemini isn't available for their account, Google suggests confirming that Gemini Apps is turned on in Family Link, waiting, because "it may take a few hours for the change to take effect," and restarting Gemini. Google also suggests checking that your child is using an available device, language and region.

What protections apply to children using Gemini

According to Google's guide for parents:

• Harmful content blocking for everyone. Gemini "attempts to identify and block harmful content," avoiding responses that violate Google's policy guidelines, such as violence, sexually explicit material and harassment.
• Extra filters for minors. Minors who are signed into their Google Account get added content filters, including filters to avoid Gemini "claiming to be a person or having human emotions, or role-playing as a harmful character."
• No Keep Activity for under 13s. Google says the Keep Activity setting "is not available to users under 13 (or the applicable age in your country)." Keep Activity is the Gemini setting that, for adults, controls whether chats are saved to the account and used to improve Google's AI. Our earlier tip explains it in detail: Check Whether Your AI Chats Are Used to Train the AI.
• Limits on Android voice features. If Gemini is the assistant on your child's Android device, Google says "Hey Google" and Voice Match are not available with Gemini Apps for users under 13.
• Lock screen access. Google notes that when lock screen access is turned on, Gemini can respond when the screen is locked. If you would rather it did not, check the lock screen settings on your child's device.

What happens when your child turns 13

This is the moment many parents miss. Google says that when a child turns 13 (or the applicable age in their country), Google emails both the supervising parent and the child to say the account can be updated.

• If nobody acts on the email, the existing supervision settings stay as they are. Google says supervision settings "remain the same until your child decides to update their account."
• The child can update to supervision designed for children over 13. That is still supervision, not the end of it.
• Children need parent approval to stop supervision until they turn 18, and Google says parents and children both get notified when supervision is stopped.

So a 13th birthday does not switch your controls off by itself, but it is a good day to sit down together and review which settings still make sense.

What Family Link does beyond Gemini

Family Link is not an AI tool. It is Google's general supervision system, and Google says it also lets parents choose which apps from Google Play a child uses, set up approvals for purchases, filter Google search results, manage browsing activity in Chrome, and see where a child is on Android devices. If a child is installing other AI apps, the Google Play approval setting is where you would notice.

Gemini on smart speakers and displays at home

If you have Google Nest speakers or displays running Gemini for Home, there is a separate set of controls in the Google Home app. Google's guide for parents says:

• "You can't use Parental controls without Family Link." First create a child account in Family Link, then add your child to your home in the Google Home app as a Member. Google says child accounts can only be Members.
• Voice Match is required for the parental controls you set for your child to work correctly.
• You can choose a content filter for voice responses on your speakers and displays. If your child under 18 has been Voice Matched and added to your home, Google says "they'll always receive filtered responses, regardless of your filter selection."
• Digital Wellbeing lets you set filters and schedule downtime, and Google says downtime "can also be used to block Gemini for Home access for your supervised children and guests."

A smart speaker in a shared room is easy to forget when you are thinking about phones. Add it to your list from Step 1.

Claude: an adults only service

Claude does not offer parental controls because Anthropic does not allow under 18s to use its consumer product. Anthropic's help centre says: "We require all users to be at least 18 years old to create and use a Claude account."

How Anthropic enforces this, according to its own pages:

• Everyone setting up an account must confirm they are 18 or over.
• Anthropic says it has "safety systems in place to detect if people under 18 may be using Claude" and will disable accounts based on indicators of minor activity. If a user says in a conversation that they are under 18, Anthropic says its classifiers flag this for review.
• A person whose account is flagged can verify their age with Yoti, a third party service, by facial age estimation, a photo of an ID document, or the Yoti Digital ID app. Anthropic says it receives only a pass or fail result.
• In certain US states, new laws require app stores to share age information with app developers, and Anthropic says the Claude mobile app uses this to enforce its minimum age. Anthropic notes that if an account is part of a family group, such as Google Family Link or Apple Family Sharing, the age on it may have been set by a parent.

What this means for you: if your teenager has a Claude account, they are breaking its rules, and the account may be disabled. There is no teen mode to switch on. The right conversation is about why, and which supervised alternative, if any, your family is comfortable with.

Common mistakes

• Sending the ChatGPT invite and forgetting it. Nothing is linked until your teen accepts. Check your Parental controls page a day later to confirm the link is active.
• Assuming one setting covers every app. ChatGPT's controls live in ChatGPT. Gemini's live in Family Link. Gemini for Home has its own settings in the Google Home app. None of them cover the others.
• Forgetting the browser. A child who cannot open an app may simply open a website instead. Family Link's Chrome controls and your conversation at home matter as much as the app settings.
• Turning off the strongest protection. OpenAI lets parents turn off the reduced sensitive content setting for linked teens. Think carefully before you do.
• Missing alerts because of an old email address. ChatGPT safety notifications go to the contact details on the parent account. Update them now.
• Treating the controls as a substitute for talking. Every company quoted here says its protections are imperfect. Your child needs to know what to do when the chatbot says something odd, upsetting or wrong.

Related checks worth doing at the same time

1. Model training. Decide whether your teen's chats can be used to train AI models. In ChatGPT you can opt them out from the parental controls page. For your own accounts, see Check Whether Your AI Chats Are Used to Train the AI.
2. Personal details. Google advises telling children not to share "sensitive personal information like their address, school name, or family information" with Gemini. The same rule is sensible for any chatbot.
3. Reporting bad answers. Google says you can report a harmful response by submitting feedback, or your child can mark it as a bad response. Show your child where that button is.
4. Voice cloning scams. AI can also be used against families from the outside. A shared family safe word is a simple defence: Agree a Family Safe Word Before a Cloned Voice Calls You.
5. Check fake images. If your child shows you a surprising picture they saw online, our tip on checking whether a photo or video was made with Google AI shows one quick way to test it.

A worked example

Here is how the steps above might play out for an imaginary family. The details of the family are made up; the settings are the ones described by the companies.

Sam is 14 and uses ChatGPT for homework. Sam's younger sister, Ria, is 9 and has an Android tablet supervised with Family Link. The family has a Nest Hub in the kitchen.

Sunday evening, step one. Their parent asks both children which AI tools they use. Sam mentions ChatGPT and says a friend uses Claude. Ria says she has asked the kitchen speaker questions.

ChatGPT for Sam. The parent opens ChatGPT on their own account, goes to Parental controls in settings and sends Sam an invite. Sam accepts on his phone. On the control page, the parent leaves Reduce sensitive content switched on, sets quiet hours for school nights, turns off image generation for now, and opts out of model training. They leave voice mode on because Sam uses it to practise a language. They agree together that Sam will say if he sees anything that worries him, and the parent explains that if Sam unlinks the account, the parent will get a notification.

Claude. The parent explains that Claude requires users to be 18 or over, so it is not an option for Sam at 14, and that accounts belonging to under 18s can be disabled.

Gemini for Ria. The parent opens Family Link, selects Ria, and goes to Controls, then Gemini, then Gemini Apps. Because Ria is 9 and the parent wants to wait, they leave Gemini Apps off. Later, if they switch it on, they know they will get an email when Ria first activates the app, and that Google's extra filters for minors will apply.

The kitchen speaker. In the Google Home app, the parent checks that Ria is added to the home as a Member and is Voice Matched, so she always gets filtered responses, and sets a downtime schedule for the evening.

Total time: about twenty minutes, most of it spent talking rather than tapping.

Keep the conversation going

Settings are a one time job. The conversation is not. Every few months, ask which AI tools your child is using now, check whether the controls are still linked and still match their age, and remind them of Google's simple advice that a chatbot "should not be treated as a friend or used to discuss personal problems or secrets." If something a chatbot says ever upsets your child, the most useful thing they can know is that they can come to you without getting in trouble.

Frequently asked questions

Can I read my teen's ChatGPT conversations with parental controls?

OpenAI's announcement of parental controls lists the settings parents can change and the safety notifications they can receive. It does not describe a way for parents to read chats, so do not count on seeing them. Talk to your teen instead.

My child turned 13. Does Family Link supervision of Gemini stop automatically?

No. Google says supervision settings stay the same until the account is updated, children need parent approval to stop supervision until they turn 18, and both parent and child are notified when supervision is stopped.

Can my teenager use Claude?

No. Anthropic says all users must be at least 18 to create and use a Claude account. Accounts it believes belong to under 18s can be disabled until the person verifies their age.

Sources

1. OpenAI: Introducing parental controls
2. Gemini Apps Help: Guide your child's Gemini Apps experience
3. Gemini Apps Help: Manage and delete your activity in Gemini Apps
4. Gemini Apps Help: What you need to sign in to Gemini Apps
5. Google For Families Help: How Google Accounts work when children turn 13 (or the applicable age in your country)
6. Google For Families Help: Manage your child's Google Account with Family Link
7. Google Home and Nest Help: Guide your child's Gemini for Home experience
8. Claude Help Center: Age assurance on Claude
9. Claude Help Center: Minimum age requirement access restriction
10. Anthropic: Protecting the wellbeing of our users