Character.AI Parental Controls: What Changed
Most articles about Character.AI and children are describing a product that no longer exists. On 25 November 2025, Character.AI removed open-ended chat for all users under 18 - the thing the platform was known for, and the thing every parenting article was warning about.
If your concern is "my 14-year-old is having long emotional conversations with an AI character on Character.AI", that specific concern was addressed by the platform itself. The more useful question now is what replaced it, and where the users who wanted that went.
What the November 2025 change did
Character.AI ended open-ended, free-form conversation for accounts identified as under 18. Under-18 users did not lose access to the platform outright; they lost the unbounded back-and-forth chat that made emotional dependency possible in the first place.
This followed sustained pressure - lawsuits, regulatory attention, and reporting on teenagers forming intense attachments to chatbot characters. It is a genuinely significant change, and it is rare: a platform removing its core feature for a third of its audience rather than adding a settings toggle.
The practical consequence for parents is that Character.AI is no longer the highest-risk AI app your child could be using. It has moved down the list.
What it did not fix
Three things, and they are the reason this is not the end of the conversation.
Age is self-declared. The restriction applies to accounts identified as under 18. A teenager who entered an adult birthdate at signup - possibly years ago, possibly without much thought - has an adult account, and adult accounts kept open-ended chat. There is no robust verification standing behind this.
The users went somewhere. This is the part that matters most. A restriction on one platform does not remove the demand; it relocates it. The apps that absorbed those users are smaller, have no safety team, no age enforcement worth the name, and in several cases no content filtering at all. A child who moved from Character.AI to one of those is in a worse position than before the ban, not a better one.
There are no parental controls in the sense you mean. Character.AI does not offer account linking, activity reports, or a way for a parent to see conversations. The under-18 restriction is a platform-level policy, not a control panel you configure. If you searched for "Character.AI parental controls" expecting a settings page, there is not one.
So what should a parent actually check
Given the above, the useful checks are different from what they would have been a year ago.
Check the birthdate on the account, not the settings. This is the whole game now. If your child's Character.AI account says they are 22, none of the under-18 protections apply to them, and correcting that does more than any external tool. Ask directly; it is usually a shrug rather than a confrontation, because most children entered a fake age for the same reason they always have - to get past a gate, not to hide anything specific.
Find out which apps are actually installed. Character.AI is a known name with a safety team under legal pressure. The apps worth your attention are the ones you have not heard of. Look through the phone and the browser history for chatbot apps generally rather than for this one specifically.
Ask what it was for. A child using AI characters for collaborative fiction is doing something creative and fairly harmless. A child using one because it is the only place they feel heard is telling you something important, and the app is the symptom. These look identical from a usage report and completely different in conversation.
The AI companion landscape as it stands
| Platform | Under-18 policy | Parent controls | Content filtering | Risk level for a teen |
|---|---|---|---|---|
| Character.AI | Open-ended chat banned since Nov 2025 | None | Strong, post-ban | Moderate, if age is honest |
| ChatGPT | Teen protections automatic since Aug 2026 | Yes, no transcripts | Strong | Low to moderate |
| Replika | 18+ | None | Limited | High for a minor |
| Chai | Nominally 18+, weakly enforced | None | Minimal | High |
| Janitor AI | Nominally 18+, weakly enforced | None | Effectively none | High |
| Google Gemini | 13+ with Family Link | Yes, via Family Link | Strong | Low to moderate |
The shape of this table is the actual finding. The named platforms with legal exposure have moved quickly on child safety. The risk has concentrated almost entirely into apps with no such exposure, which is exactly where blocklist-based filtering performs worst, because a filter has to know an app exists before it can stop it.
Why emotional dependency is the risk that matters
Explicit content gets the headlines, and it is the easier problem - it is detectable, and platforms under pressure can filter it.
Dependency is harder because nothing looks wrong. A chatbot is available at 2am, never tired, never distracted, never has its own bad day, and is built to keep the conversation going. For a child who is lonely, anxious or socially struggling, that is not a novelty - it is a relief. And it displaces exactly the difficult, awkward human interactions that build the ability to handle difficult, awkward human interactions.
The signals worth watching are ordinary ones: withdrawal from friends they used to see, secrecy specifically around one app, distress when they cannot access it, and a device that comes to bed. The signs of phone dependency generally apply almost unchanged here.
If you see those, the technical response is secondary. A child who has substituted an AI for people needs the people part addressed; blocking the app on its own removes the relief without replacing it.
Where 3Eyes Fits
The problem this page keeps arriving at is that the risk is in apps you have not heard of yet. That is a structural mismatch with blocklists: every new companion app has to be discovered, named and added before it can be stopped, and new ones launch continuously.
3Eyes inverts it. Nothing opens unless you have approved it, so an AI companion app that launched last week is blocked because it is not on your list, not because someone catalogued it. You can allow chatgpt.com for homework and still have every unreviewed chatbot site blocked by default. It runs on the computer rather than the network, so it holds on a hotspot and at a friend's house, and it needs your password to uninstall.
For the dependency side specifically, the more useful half is what else is on the machine: 3Eyes carries educational games, learning activities and a built-in tutor, and restricts YouTube to channels you approve rather than blocking it. Setup is about two minutes. See how it works.
Frequently asked questions
Does Character.AI have parental controls? Not in the conventional sense - there is no account linking, no activity report and no way to read conversations. What exists is a platform-wide policy removing open-ended chat for under-18 accounts, applied automatically rather than configured by a parent.
Can my child get around the under-18 restriction? By having entered an adult birthdate, yes, and many did so long before this was a concern. Checking the age on the account is the highest-value thing a parent can do here.
Is Character.AI safe now? Safer than it was, for accounts that honestly report an under-18 age. The residual risks are the fake-age problem and the fact that children who wanted unrestricted AI companionship largely moved to apps with far weaker safeguards.
What should I do if my child is emotionally attached to an AI character? Treat it as information about their social life rather than as an app problem. Abruptly removing the app addresses the symptom and can make a lonely child lonelier. Ask what they talk about and what it gives them, and work on the human side alongside any limits you set.
Which AI apps should I be most concerned about? Not the famous ones. The apps with no safety team, no age enforcement and no filtering are where the genuine risk sits, and they are precisely the ones you will not recognise on a screen-time report.