AI chatbot settings and parental controls: a checklist for parents
Before a child or teen uses an AI chatbot, check six kinds of settings: the minimum age in its terms, whether chats are saved, whether chats are used to train the AI, content filters, what a parent can see, and purchases. Then add device controls for time, apps and spending. No setting replaces talking together.
Last reviewed: 20 October 2026 9 min read
What should a parent check before a child uses an AI chatbot?
Look for the same seven things in any chatbot or app with a chatbot inside it. The names of the switches differ from app to app; the jobs are the same.
| # | What to check | What to look for | Why it matters |
|---|---|---|---|
| 1 | Age rules | The minimum age in the terms, and whether a teen needs a parent's permission | Some makers set 13, some set 18 |
| 2 | Saved chats and memory | Whether chats are kept, whether you can delete them, and whether the app remembers details between chats | Saved chats can be kept for some time, even after you change a setting |
| 3 | Use for training | A switch that decides whether chats help improve or train the AI | On some apps this is on until someone turns it off |
| 4 | Content filters and features | A setting that reduces sensitive content, and switches for voice and image creation | Fewer features means fewer surprises for a younger user |
| 5 | What a parent can see | Whether a parent account can be linked, and what it shows | Linking may change settings without showing any chats |
| 6 | Time limits | Quiet hours inside the app, or limits set on the device | Time rules work best when they live in one place |
| 7 | Purchases | Paid plans or extras inside the app, and who approves them | Device purchase approval catches what the app does not |
The sections below take each one in turn. Every setting named here was checked on a maker's own help page when this guide was last reviewed. Apps change their settings often, so check the app itself before you rely on any one switch.
What are the age rules for AI chatbots?
There is no single age. Each maker sets its own in its terms of use.
- One widely used assistant's terms say a user must be at least 13, and anyone under 18 needs a parent's or legal guardian's permission.
- Another assistant's consumer terms say a user must be at least 18.
- A third maker lets a parent turn its assistant on or off for a supervised child's account, using its family account app.
Why 13 comes up so often: under the US children's privacy rule (COPPA), covered sites and apps must get a parent's consent before collecting personal information from a child under 13.
What to do: read the minimum age in the terms before your child signs up, not after. If the app is below your child's age, the simplest setting is "not yet".
Can a parent see a child's chatbot conversations?
Usually not, and it is worth knowing that before you link accounts.
On one major assistant, a parent can link their account to a teen's account and manage settings. Its help page says linking does not give the parent access to the teen's conversations, chat history or real-time activity. The parent may get a safety notification in limited, serious situations, and the help page says these notifications are not real-time monitoring. Either the parent or the teen can unlink the accounts, and the parent is told if the teen unlinks.
So parental controls on a chatbot are mostly settings, not a window into the chats. If you want to know what your child asks a chatbot, the reliable way is to ask, or to sit nearby while they use it.
Which chatbot settings can a parent change?
On the assistant above, a linked parent can change settings of these types, as its help page lists them:
| Type of setting | What it does |
|---|---|
| Sensitive-content reduction | Adds safeguards that help reduce sensitive or age-inappropriate content |
| Use for improving the AI | Decides whether the teen's chats can be used to help improve the models |
| Saved memories | Decides whether the assistant uses saved details to personalize answers |
| Voice | Turns voice conversations on or off |
| Image creation | Turns creating and editing images on or off |
| Study-mode start | Makes new chats start in a step-by-step study mode |
| Quiet hours | Sets times when the teen cannot use the assistant |
One detail matters for most families: on that assistant, the switch that lets a teen's chats help improve the models is on by default after linking. A parent who does not want those chats used has to turn it off.
Other apps may offer fewer of these, or none. If an app has no parent settings at all, the device controls below still apply.
What happens to the chats a child types?
It depends on the app's activity or history setting. One maker's help page explains its own version clearly:
- When the activity setting is on, chats are saved to the account and used to improve the maker's services, including training its AI, and some chats are read by human reviewers.
- When it is off, new chats are not used for training unless the user sends feedback. They are still kept for 72 hours.
- Deleting activity does not delete chats that reviewers already looked at. Those are kept for up to three years.
The same maker's help page for children's accounts says this activity setting is unavailable to users under 13.
What to do: teach one plain rule. Never type anything into a chatbot that you would not want a stranger to read: a full name, home address, school, passwords, photos of faces, or private worries. If a worry is real, bring it to a person.
What device controls help with AI apps?
Device controls work across every app, so they catch what an app's own settings miss. Phones, tablets and computers from the large makers include tools of these types:
- Content and privacy restrictions. Choose age ratings for apps and other content, and manage which privacy settings are allowed.
- Purchase approval. A child asks before downloading or buying an app, and a parent approves or declines the request.
- Screen-time limits. Set limits, and set a device to lock after a certain amount of use.
- App store content restrictions. Limit downloads and purchases by maturity level. One maker's help page warns that these restrictions don't prevent restricted content from showing up in search results or through a direct link.
- App permissions. Apps may ask for your location, contacts or photos. Turn off permissions an app doesn't need, or delete apps that ask for far more than they need.
A chatbot inside a web browser skips the app store, so browser and website settings matter too.
How do you set up an AI chatbot for a child, step by step?
- Read the terms. Find the minimum age and the rules for users under 18.
- Decide together. Agree why your child wants to use it: homework help, curiosity, or something else.
- Use a parent link if the app offers one. Link your account to your child's account.
- Go through the seven checks. Turn on content reduction, turn off training use if you prefer, and switch off voice or image creation for younger users.
- Add device controls. Set purchase approval, time limits and app permissions.
- Write it down. A family media plan puts the rules in one place. The American Academy of Pediatrics offers a plan maker that covers times and places without screens and using parental controls on devices, routers and accounts.
- Check again in a few months. Apps add features. A switch that did not exist when you set things up may be on by default now.
What can't parental controls do?
- They can't make answers true. One maker's own help page says its assistant "can hallucinate and present inaccurate information as factual." A US government AI risk guide describes the same problem: confidently stated but false content.
- They can't catch everything. Filters reduce content; they do not block all of it.
- They can't read the chats for you. On the assistant described above, safety notifications are limited and are not monitoring.
- They can't replace a conversation. The most useful control is a child who knows they can tell you when something online felt wrong.
What goes in a family agreement about AI chatbots?
Keep it short enough to fit on the fridge. Five questions cover most of it:
- Which chatbots, and for what? (For example: "for questions, not for writing my homework for me.")
- Where? (For example: in shared rooms, not behind a closed door.)
- What never goes in? (Names, addresses, school, passwords, faces, secrets.)
- What do we do with an answer? (Check it with a second source before using it.)
- What happens when something feels wrong? (Stop, close it, tell a grown-up. Nobody gets in trouble for telling.)
Where does Mechna Academy fit?
Mechna Academy is a city of machines for curious children aged 8 to 18, opening first to ages 8 to 10. The machines children study are simulations built from rules and examples written and checked in advance. We never use children's data to train AI.
That means there is no chatbot to set up. Children learn how AI works by examining machines, rather than by talking to one.
Sources
- Terms of use, minimum age
- Consumer terms, minimum age
- Children’s account settings
- Protecting your child’s privacy online
- Parent account controls
- Chat activity settings
- Content and privacy restrictions
- Purchase approval
- Screen time limits
- App store content restrictions
- App permissions
- Family media plan
- AI risk guide
Questions
It depends on the chatbot. Each maker sets a minimum age in its terms: one widely used assistant says 13, with a parent's permission under 18, and another says 18. Read the terms before your child signs up.
Usually not. On one major assistant, linking a parent account lets the parent change settings, but its help page says it does not give access to the teen's conversations or chat history. Asking your child, or sitting nearby, is the reliable way to know.
Some can, unless a setting is turned off. One assistant's parent setting that lets a teen's chats help improve its models is on by default, and another maker's activity setting, when on, uses chats to train its AI. Check the training or activity setting in each app.
No. Controls change settings and reduce some content, but they cannot make answers true or catch everything. Pair them with device limits, a short family agreement and regular conversations.
No. The machines children study are simulations built from rules and examples written and checked in advance.
We never use children's data to train AI.
Mechna Academy is a city of machines for curious children aged 8 to 18, opening first to ages 8 to 10.
Start for freeThe first missions, for ages 8 to 10, are free; membership comes after. Places are limited: we email you when your family's place opens.