Character.AI Face Scan 2026: Why It Happens, What to Do

The prompt lands mid-scene, which is the part that stings. You're four messages into a conversation, nothing weird is happening, and Character.AI decides it isn't convinced you're an adult. Then it asks for a camera scan.

If you typed character ai face scan into Google with your jaw somewhere near the floor, this is the user-side explainer. Not the parent version. What set the flag off, where the selfie goes, whether you can clear it without handing anyone a driver's license, and what the app strips out if you walk away. I've watched Character.AI tighten one screw after another since early 2025. This is the tightest one so far.

What sets off Character.AI age assurance

The thing most people get backwards: the camera isn't the age check. The age check already happened, silently, and you lost. Character.AI's own Age Assurance FAQ says the system runs automatically on all accounts and weighs three things: “login info, activity on the platform, and some signals from third parties.” That last bucket is doing a lot of quiet work, and the company doesn't itemize it.

Only when that scoring guesses minor does anything visible happen. So the accounts getting caught aren't the suspicious ones. They're the thin ones. A brand-new sign-up. An account with eleven messages of history. An email address nothing in the third-party data can corroborate. A year of chat logs and a card on file, and you'll probably never see a camera prompt.

Rollout is also gradual, which explains the most common confusion I see in comment threads. Your friend uses the same app on the same phone and has never been asked. That doesn't mean you did something wrong. It means the two of you are in different waves.

If you're new to the platform generally, the complete Character.AI guide covers how the account system fits together, and our safety breakdown covers the older moderation layers this sits on top of.

My main account was never asked. A fresh one was.

I keep throwaway accounts around for testing, which is how I ended up sitting in front of this thing on September 4, 2026. I'd spun up a clean account to look at new-user onboarding, typed a few unremarkable messages, and got the under-18 screen almost immediately.

My irritation lasted ten seconds before it turned into irritation at myself. Of course the fresh account got flagged. It had no history, nothing for the system to weigh except an email nobody has ever heard of. Meanwhile my real account, the one where I built Teodora back in July, has never been prompted once. It passed silently, on behavior, exactly the way the FAQ says it should.

The scan was faster than I expected. Tap through, camera opens, an oval appears, you center your face and hold still for a beat. It reminded me of an airport passport kiosk, minus the line and the guy behind me sighing. Green check, back in the app, under a minute. No ID request, no follow-up email, no “we're reviewing your submission” purgatory.

Where the selfie goes, and what Character.AI never sees

The Character.AI selfie verification flow hands off to Persona, an identity verification company that also runs checks for LinkedIn, OpenAI, Block, and Etsy. That list matters. This isn't a vendor stood up last quarter to satisfy a lawsuit, it's infrastructure large companies already lean on.

Two claims carry all the weight here, and both come straight from the FAQ. Character.AI says it never receives your selfie or your ID, only the pass/fail result. And any biometric information collected during the process is deleted within seven days, with IDs held by Persona for one week.

I like that architecture. A pass/fail token is a far smaller thing to leak than a face database, and it beats what I've seen on smaller platforms, which is emailing a photo of your license to whatever support address answers first. The AI companion privacy checklist walks through what these apps retain, and the 2026 privacy guide covers who your chat data gets shared with once it's stored.

What I can't tell you is whether the deletion happens. Nobody outside Persona can verify a seven-day retention window. It's a policy statement, not an audit, and you should file it in your head as such. One more line worth knowing: Character.AI has said that if it ever launches a virtual currency, it may create “very limited exceptions” where it works with Persona to access an ID for fraud resolution, with notice and consent. That door is closed today. They built a hinge on it anyway.

When the selfie fails and Persona asks for ID

ID is not the default. It only comes out when the selfie result is inconclusive, meaning the model couldn't confidently place you on the right side of 18. Then you get the document step.

Accepted documents, per the Age Assurance Troubleshooting page: driver's license, passport, passport card, ID card, permanent residence card, immigration visa, consular ID, and certificate of citizenship. It has to be the physical document. A photo of a digital copy gets rejected, which trips people up when they try to shoot their phone screen showing a scanned passport.

There is no route three.

Character.AI states the selfie and the ID are the only accepted methods. If the selfie comes back inconclusive and you don't have a valid physical ID, the account can't move forward. Not a downgrade, not a wait. Stuck.

I haven't been through the ID path myself. My test account cleared on the selfie, so this section is documentation rather than experience.

How to pass Character.AI selfie verification on the first try

Almost every failure I've read about comes down to something dull and fixable. The troubleshooting page is unusually specific, and the logic behind it is that the system wants to be sure your face doesn't look like a deepfake. Anything that alters your features reads as tampering.

• Kill the filters. Beauty filters, drawn-on features, and heavy makeup effects all get flagged. It's the first thing the troubleshooting page brings up.
• Light in front, not behind. Backlighting and glare wreck the capture. Turn away from the window.
• Camera at eye level. The classic phone-in-lap angle fails a lot.
• One face in frame. Your roommate wandering past behind you counts as a second face.
• Hold still. Motion blur is a bigger killer than people assume.
• Nothing covering your face. No sunglasses, no mask, no hat brim casting a shadow.

If your laptop webcam is the weak link, you don't have to abandon the flow and start over. The screen offers a QR code that moves the session to another device mid-process, so you can finish on your phone. If you've made a few genuine attempts and it still won't clear, stop grinding and open a support ticket. Repeated failed attempts don't improve your odds.

What happens if you just refuse

You keep your account. You lose the thing you came for.

Declining verification drops you into the under-18 experience, which means no open-ended chat. Character.AI pulled that for minors back in November 2025 and replaced it with structured creative tools, Stories and AvatarFX-style features, where you build something rather than talk to someone. It's not a punishment tier exactly, but it is a different product. Our breakdown of the teen chat ban shows what that restricted account actually looks like day to day.

For a paying adult that's the genuinely bad outcome: still subscribed to a product you can no longer use as intended. If you land there and won't verify, cancel rather than pay for a stripped-down tier. Our Character.AI+ value breakdown covers what you're actually paying for, the alternatives roundup lists platforms with lighter identity requirements, and the SillyTavern comparison is the local-first route for people who'd rather not hand any platform a face at all. Before you go anywhere, though, run a data export so your history isn't stranded behind a wall you can't clear.

Things that don't get you around it

A Character.AI+ subscription does not bypass age assurance. I want to be unambiguous, because it's the most common assumption I see and it's wrong. Your card on file, your billing address, your two-year payment history: none of it counts as proof of age. Neither does a Google account with your birthday sitting right there in it.

No utility bill option, no credit check option, no “email a photo of your student card to support” option either.

Wrongly flagged? Trigger Character.AI age verification yourself

If you got scored into the wrong bucket and want to fix it on your own schedule, the path is Settings → Advanced → Verify Age. It's on the web and in the mobile apps, but the app has to be on the latest version or the option simply won't render. That trips up a lot of people who then conclude it doesn't exist for them.

Same path if you've turned 18 and want to migrate from the teen experience to the adult one, and you may be asked for the selfie at that point. If you somehow got verified into the wrong experience after passing, don't keep re-scanning. File a ticket and let a human sort it.

The legal squeeze that put a camera in your chat app

None of this came out of a product brainstorm. Character.AI banned open-ended chat for minors in November 2025, settled five lawsuits in January 2026, and then caught a fresh one on August 13, 2026 in the Northern District of California over alleged teen mental-health harm. Our earlier coverage of the company's legal trouble has the background on how that pressure built.

The regulators arrived right behind the plaintiffs. California's SB 243 put companion chatbots under specific obligations, and the federal GUARD Act, which carries an age-verification mandate, has cleared committee and now sits ahead of a Senate floor vote. The state and federal rules explainer has the detail on both.

On timing, be skeptical of anyone giving you a precise switch-flip date. The system was announced alongside the under-18 ban in late 2025 and enforcement widened gradually through 2026, with third-party reports putting broad enforcement somewhere in the first half of the year. The official pages don't name a month, and neither will I.

Here's where I'd argue with people. Everyone is upset about the selfie, and the selfie is the least invasive piece of the whole system: a forty-second check against a vendor that promises to delete it. What should bother you is the layer above it. An invisible scoring system has already formed an opinion about who you are, using data brokers you never agreed to, and you only find out when it gets you wrong. Our red flags guide and the safety audit of companion apps both apply well beyond Character.AI.

So, do you scan?

I did it on a throwaway account without much hesitation, and I'd do it on my main one. The tradeoff is a selfie held by a company that also verifies LinkedIn users, in exchange for keeping an account I use several times a week. For me that math is easy.

I'd think harder if I lived somewhere with weak data protection law and no practical way to complain when a policy gets broken. Those are real situations and I'm not going to wave them off. If that describes you, the local-first route exists for exactly this reason, and the Replika comparison shows how differently two big platforms handle identity.

What I keep coming back to is how ordinary this felt. A year ago the idea of a chat app asking for my face would have seemed absurd. On September 4 it took forty seconds and I barely thought about it. That adjustment is fast, and I'm not sure it's good.

Did your account get flagged, and what did you actually do about it? I'm collecting reports of who's getting caught, especially older accounts with real history, because that pattern would change what I think is going on. The easiest way to reach me is to join the newsletter and reply to any issue — mention how long the account had been active.

The rest of your Character.AI face scan questions