Discover the step-by-step guide on how to set parental controls on AI apps and protect your children from inappropriate content.
Here’s my honest position after an evening of clicking through every one of these menus. Four of them are real. The rest are theater, and the companies building them know it.
You can do the useful parts in about twenty minutes. Phone in hand, kid’s password known, coffee nearby. I’d start with whichever app is already open on their home screen, because that’s the one doing the work.
ChatGPT: link the accounts, then kill the memory
Open Settings from your profile picture. On mobile it hides under App Settings. Tap Parental Controls, then Add family member, then pick email or phone. Choose “My child” from the relationship dropdown and send it. Your teen taps the link and the accounts connect.
Now the toggles. Reduce sensitive content is on by default. Quiet hours blocks access on a schedule — I set 10pm to 7am and stopped thinking about it. You also get switches for voice mode, image generation, study mode, and model improvement.
The one I care about is memory references. It ships on. Memory is what turns a chatbot into something that remembers your kid’s ex-girlfriend’s name and asks how that’s going. Researchers keep flagging exactly this as the feature that blurs tool into relationship. Turn it off.
What you don’t get: transcripts. No chat history, no live feed, nothing. OpenAI says employees review flagged self-harm content and aim to alert you within an hour by text, email, and in-app. If your teen unlinks the account, you’re notified. That’s the whole deal, and it’s more than most.
Gemini: three taps deep in Family Link, which is not an accident
Go to familylink.google.com or open the app. Pick your child. Tap Controls, then Gemini, then Gemini Apps, and flip the switch. Older app versions route you through Controls → Content restrictions → Gemini instead.
Under-13 accounts can’t touch Gemini until you switch it on yourself. Google emails you the first time they open it, which is a nice touch and also the last time Google tells you anything. The Keep Activity setting isn’t available to under-13s, so there’s no history for you to read even if you wanted to.
Worth knowing: a Chromebook loaned out by the school district may run a different account with different rules entirely. Check both. School Workspace accounts often have Gemini switched on by an IT admin you’ll never meet.
There’s a second door, too. On most newer Android phones, holding the power button no longer summons Assistant — it summons Gemini. That swap happened quietly, and the toggle above is what closes it.
Character.AI and the copycats nobody’s watching
Character.AI pulled open-ended chat for under-18s on November 25, 2025. They ramped it down first — two hours a day, then less, then gone. Age checks now run through an in-house model plus Persona, the third-party ID vendor. Parental Insights sends you a weekly email with daily time spent and the characters they talked to most. No transcripts there either. Setup runs backwards from what you’d expect: the teen has to send you the invite from their own settings, so you can’t do this behind their back.
So that one’s handled, more or less. My worry is the forty knockoff companion apps that flooded the App Store after Character.AI tightened up. Those have no Family Link entry, no parent dashboard, no press release. Search your kid’s phone for “AI,” sort by date installed, and look at what’s actually there.
Snapchat and Instagram: read the fine print twice
Snapchat’s Family Center lets you see who they message and restrict sensitive content. My AI, though, sits pinned at the top of their chat list and won’t leave unless you pay for Snapchat+. Your only free move is Settings → Privacy Controls → Clear Data → Clear My AI Data, repeated forever. Using AI without paying is fine for your kid and apparently a design constraint for you.
Instagram is better, barely. Parents in the US, UK, Canada, and Australia can now switch off one-on-one chats with AI characters, block individual characters, and see topic summaries. Notice the gap. Meta AI itself stays on, always, with what Meta calls age-appropriate defaults. You cannot turn it off. That tells you what the controls are for.
The phone-level backstop, which is the one that holds
App-level settings are polite requests. The operating system is the actual lock, and I’d set it even if you trust your kid completely, because chatgpt.com works fine in Safari and your app block does nothing about it.
On iPhone: Screen Time → Content & Privacy Restrictions. Under Allowed Apps & Features you can strip out Apple Intelligence. Then open Content Restrictions → Web Content, choose Limit Adult Websites, and add the chatbot domains to Never Allow by hand. Type them in. It’s tedious and it takes four minutes.
On Android: Family Link → Controls → Apps, where you set a daily limit or block the app outright. Pair it with site blocking in Chrome under the same menu.
Set the App Store and Play Store age rating while you’re in there. Most of these apps carry a 12+ or 17+ label that does real work if you let it.
What I’d skip, and the part settings can’t do
Skip the monitoring apps promising to read AI conversations. They can’t. The transcripts don’t leave the platform, and the vendors selling you otherwise are selling screenshots of a keyboard logger that breaks on the next iOS update.
Skip the impulse to block everything. Homework help, recipe scaling, debugging a Minecraft mod — AI for small tasks is genuinely good, and everyday AI tools are now as ordinary as a calculator. Blanket bans just teach kids to use a friend’s phone.
The conversation I’d actually have is about when not to use AI. Not as a lecture. As a running joke, ideally. My rule at home: if the answer matters and you can’t check it, don’t ask a chatbot. Sixteen-year-olds are excellent at spotting AI hype vs reality once you give them permission to be cynical.
One day they’ll ask the real question: is AI worth it? Answer honestly. Sometimes, for some things, and almost never for the things that hurt. Say that part out loud, because no toggle will say it for you.
A twenty-minute checklist, because you’ll forget otherwise
Do these in order and you’re done:
- Link ChatGPT parent and teen accounts. Memory off, quiet hours on.
- Family Link → Controls → Gemini. Decide on or off, deliberately.
- Check Character.AI’s Parental Insights email. Then audit the App Store for knockoffs.
- Instagram: AI character chats off. Snapchat: clear My AI data, accept the limits.
- Screen Time or Family Link at the OS level. Add the web domains manually.
- Set store age ratings on both app stores.
Recheck in three months. These products ship changes faster than anyone can write about them, and half the settings above didn’t exist eighteen months ago. Mine already moved once while I was writing this.
One last thing. Every control here depends on knowing which apps your kid actually uses, and that information does not come from a settings menu. It comes from looking at the phone. Ask first, obviously — but look.
Sources: OpenAI Help Center · Engadget · Google For Families Help · Character.AI blog · Meta Newsroom · Internet Matters
