Character AI Age Verification: What Changed and What To Do

What changed in April 2026

Before 2026, Character.AI operated on self-declared age. You entered a birthdate at signup and the platform largely took your word for it. That model collapsed under legal pressure. As of April 2026, age assurance is mandatory: every account has to be resolved as adult or minor, and accounts that cannot be resolved are restricted rather than left alone.

The practical consequences for an existing account are narrower than the headlines suggest, but they are real:

• You cannot opt out. Ignoring the prompt is treated as a failed check, not as a deferral.
• Accounts that fail or stall are pushed into a restricted mode. Users describe it as effectively read-only — you can see history, but open-ended chat is limited or blocked.
• Verification is handled by Persona, a third-party identity provider, not by Character.AI directly. Your face image and any ID go to Persona’s pipeline.
• Passing once does not always stick. A minority of users report being re-prompted after a device change or a long gap between sessions.
• Under-18 accounts are not simply verified and released — they get a materially different, more restricted product.

How the age check actually works, step by step

It is a cascade, not a single test. Most accounts never see a camera. Understanding which stage you are stuck at is the difference between a two-minute fix and a two-week support thread.

1. Signal-based estimation. Character.AI first infers your likely age band from account signals — how long the account has existed, the declared birthdate, and behavioural patterns on the platform. A long-standing adult account usually clears here silently and never sees a prompt.
2. Selfie age estimation via Persona. If the model is not confident, you are handed to Persona and asked to take a live selfie in-app. This is age estimation from a face, not facial recognition against a database — the model returns an estimated age band, and if that band sits safely above 18 you are through.
3. Government ID fallback. If the selfie estimate lands near the threshold or fails repeatedly, the flow escalates to a document upload — typically a passport, driving licence or national ID — often with a second selfie to match against the document photo.
4. Restriction or release. Clear any stage and your account returns to normal, usually within minutes. Fail all of them and the account stays restricted until an appeal succeeds.

Why Character.AI added age verification

This was not a product decision made in a vacuum. Character.AI was named in minor-safety lawsuits filed in 2024 and 2025 alleging that its chatbots caused serious harm to teenage users, including cases involving self-harm. Those suits ran alongside a broad regulatory shift — state-level age-assurance laws in the US, the UK Online Safety Act’s children’s codes, and the EU’s Digital Services Act all push in the same direction: platforms that host open-ended, emotionally intense conversation are increasingly expected to know whether they are talking to a child.

Whether you think the change is proportionate depends largely on what you used Character.AI for. If you were an adult writing long collaborative fiction, an identity check for a chat app feels heavy-handed. If you are a parent who found an emotionally dependent thirteen-year-old mid-roleplay, it looks overdue. Both reactions are reasonable, and the platform has to serve both.

What to do if you are wrongly locked out

There is no instant override and no phone line. What works is making the automated check succeed, in this order — most of the reported false negatives are environmental, not permanent.

1. Retry the selfie on a phone, not a laptop webcam. Front-facing phone cameras are substantially better, and the estimation model is sensitive to resolution.
2. Fix the lighting before you fix anything else. Face a window or a lamp; avoid backlighting, heavy shadow, and overhead-only light. This single change resolves a large share of failures.
3. Remove glasses, hats and masks, tie hair back, and keep a neutral expression. Heavy makeup and beauty filters both skew age estimates — turn off any camera enhancement.
4. Do not retry ten times in a row. Repeated rapid attempts can trip rate limits and make the case harder to appeal. Two or three clean attempts, then stop.
5. Escalate to the ID step deliberately if the selfie keeps failing. A passport or driving licence resolves the question definitively, and it is usually faster than an appeal.
6. File a support appeal with your Persona case or inquiry reference. Keep the reference number — a ticket without it is very hard for support to trace.
7. Give it time and do not open duplicate tickets. Duplicates typically reset your place in the queue rather than speeding it up.

What not to bother trying

• Editing your birthdate. Changing a declared birthdate after being flagged is usually blocked, and where it is not, it tends to be read as an evasion signal.
• A VPN. Age assurance is tied to the account, not to your country. A VPN changes nothing here and can add a fraud flag.
• Someone else’s face or ID. This is straightforward identity misuse, it breaks Character.AI’s terms, and it is the one route that turns a temporary restriction into a permanent ban. We are not going to help you do it.
• Deleting and re-registering. A new account starts with no history and no positive age signals, which makes you more likely to be flagged, not less — and you lose everything.

Privacy: what a face scan actually involves

This is the part worth reading properly before you decide, because "face scan" covers two very different things. Age estimation analyses an image to guess an age band and does not need to know who you are. Identity verification matches a document to a face and by definition does. Character.AI’s flow starts with the first and escalates to the second, which means the privacy question changes depending on how far down the cascade you go.

• Selfie age estimation does not identify you by name. It returns an estimated age, and vendors generally state that the image is deleted after processing — but "generally" is doing work in that sentence, so read the retention policy shown in the flow.
• The ID step is different in kind. Uploading a passport or licence means a third party now holds a government document tied to your face and your use of a companion-chat app.
• Persona is a mainstream identity vendor used across fintech and marketplaces, not a fly-by-night operation. That is reassuring about competence, not a guarantee about retention.
• Biometric data is regulated unevenly. Illinois’ BIPA, Texas’ CUBI, the UK and EU GDPR regimes and several other jurisdictions give you rights over biometric processing — including, in many cases, the right to request deletion.
• If handing over an ID to chat with a chatbot is the line for you, that is a coherent position. The honest answer in that case is to use a different app, not to look for a workaround.

Age rating versus age verification — they are not the same thing

These get conflated constantly in forum threads. An age rating is a storefront label: Apple and Google assign one based on what an app may contain, and it gates nothing beyond a parental-controls setting on the device. Age verification is an in-product check that decides what your specific account can do. An app can carry a 17+ or 18+ rating and never verify anyone; an app can carry a 12+ rating and verify everyone. Character.AI moved from the first posture to the second, which is why long-time users experienced it as a sudden change rather than a gradual one.

If you would rather not verify: the realistic alternatives

Be clear-eyed about the trade before you move. Nothing else has Character.AI’s library of user-made characters — that library is the product, and it is not reproducible elsewhere. There is also no export path: you cannot take your characters or your chat history with you, so switching means starting over. If the character roster is what you value, verifying is genuinely the better option.

If what you actually wanted was one companion who remembers you, rather than a cast to write with, the calculus is different. Joii is our own app and sits in that second category: a single safe-for-work AI companion for iPhone with persistent memory, no romance mode and no character roster. It is iOS-only, and if you came to Character.AI for collaborative fiction with a hundred different characters, it will disappoint you. Kindroid and Nomi are stronger picks for character depth and long-term memory; Talkie is the closest free substitute for browsing a character library.

Written by The Joii Editorial Team. We write about AI companions, loneliness and digital wellbeing. We are not clinicians, and nothing here is medical advice.