AI chatbot age verification: which apps check, how, and where

AI companion apps check age with a birthday or 18+ confirmation at sign-up, an age estimate, or a selfie or ID check run by an outside verifier. The stronger checks mostly apply where a law requires them, such as in the UK, Australia and US states with adult-content age laws. Character.AI estimates age on every account and Kindroid on every account without a paid subscription; both ask for a selfie or ID when unsure.

Key facts, with the source and the date each was checked
US Supreme CourtFree Speech Coalition v. Paxton, decided June 2025: Texas's law requiring sites with sexual material harmful to minors to verify that visitors are 18 survives intermediate scrutiny · supremecourt.gov, checked
UKOfcom's age assurance guidance for online pornography was published in January 2025, and the Online Safety Act's child safety duties were in full effect by summer 2025 · gov.uk, checked
First UK case against a companion siteOfcom investigated Novi Limited, operator of joi.com, over age checks, then closed the case without findings after it added age assurance for UK users · ofcom.org.uk, checked
New YorkChapter 368 of 2026 (S 9051B), signed October 28, 2026 and in force January 1, 2027: chatbot operators may offer 'unsafe AI companion features' (claiming to be human or have emotions, a personal relationship, flattery, memory of personal details beyond 12 hours, sexually explicit content and others) only to users an age check has determined are not minors, must offer more than one check method, at least one not relying on government ID or keeping the user anonymous to the operator, and must delete what a check collects immediately after it · nyassembly.gov, checked
AustraliaIndustry codes in force since March 2026 require AI companions that can generate sexually explicit or self-harm material to confirm users are 18 or older, using checks such as ID, facial estimation or a credit card · esafety.gov.au, checked
US map of states with a law covering AI companion or therapy chatbots: 14 in force, 3 from January 2027, 4 from July 2027, 30 with no such law
Each US state by when its first law covering AI companion or therapy chatbots takes effect, from our AI companion law tracker. Free to reuse with credit: link to this page. Full-size PNG

If an AI companion app has just asked you for a selfie or a photo of your ID, you are not imagining a new trend. For years the only age check in this category was a birthday field. Since 2025 a mix of court rulings, national online safety laws and companion-specific bills has pushed apps to check properly, at least for adult content and at least in some places. This guide explains which apps check, where, how the checks work and what they cost you in privacy.

Which apps check age, and how

The table shows what each companion app documents about age checks, from its own terms, help pages or verifier. Follow an app's name for its review, where the policy is quoted with its date.

How each AI companion app checks age, from its own documents
AppMinimum ageHow age is checkedAdult content
AIGirlfriends.ai18+Self-declaredAllowed for adults
AnimaNot statedNot documentedNot stated
Candy.ai18+Self-declaredAllowed for adults
Chai AI18+Third-party check (face or ID)Allowed for adults
CrushOn AI18+Self-declaredAllowed for adults
DarLink AI18+Self-declaredAllowed for adults
DreamBF18+Self-declaredAllowed for adults
DreamGF18+Self-declared; selfie or ID check in some placesAllowed for adults
FantasyGF18+Self-declaredAllowed for adults
GirlfriendGPT18+Self-declaredAllowed for adults
HeraHaven18+Self-declaredAllowed for adults
Janitor AI18+Self-declared; selfie or ID check in some placesAllowed for adults
Kindroid18+Age estimation, then a selfie or ID check if unsureAllowed for adults
Kupid AI18+Self-declaredAllowed for adults
Muah AI18+Self-declaredAllowed for adults
My Dream Boy18+Self-declaredAllowed for adults
Nectar AI18+Self-declaredAllowed for adults
Nomi.ai18+Self-declaredAllowed for adults
OurDream AI18+Self-declared; selfie or ID check in some placesAllowed for adults
SpicyChat AI18+Self-declared; selfie or ID check in some placesAllowed for adults
Animates18+Not documentedOff by default; adults can switch it on
Intimate18+Self-declaredOff by default; adults can switch it on
Zeta13+Self-declaredOff by default; adults can switch it on
Character AI13+Age estimation, then a selfie or ID check if unsureNot allowed
EVA AI18+Self-declaredNot allowed
PolyBuzz18+Self-declared; selfie or ID check in some placesNot allowed
Replika18+Self-declaredNot allowed
Talkie16+Self-declaredNot allowed
Tipsy Chat18+Self-declaredNot allowed

Where an app checks more strictly in some places, its row says so. Follow the app's name for the countries and states, the verifier it uses and the date we read its policy.

The four kinds of check

Self-declaration. You type a birthday or confirm in a tick box that you are 18. Nothing is uploaded, so there is no document or face image to store. Regulators in the UK and Australia say it is not enough on its own for the services their laws cover, which is why apps add one of the checks below in those places.

Age estimation. The app guesses your age without documents, either from a face scan or from how you use it. Character.AI runs an in-house model on every account, reading login details, activity and some signals from third parties, and sends cases it is unsure about to a third-party verifier, Persona, for a selfie check, with an ID upload only if the selfie leaves Persona unsure. Character.AI says it never receives the ID and Persona keeps it for one week. An adult placed in the under-18 version can start the check at any time under Settings → Advanced → Verify Age, on the web or in the app. Kindroid counts a paid subscription as proof of age. Other accounts get a behavioural age estimate, plus the age range the App Store or Google Play shares with its apps, and a flagged account must subscribe or pass a Yoti check within 30 days or be deleted. Since Kindroid's November 2026 update, Yoti's selfie passes only if it estimates you are 21 or older; 18 to 20 year olds use an ID. Most adults never see a prompt.

Face or selfie check. A short video selfie is analysed to estimate whether you are over 18. It needs no document, and the verifier's privacy notice says how long the image is kept. This is the method most apps offer first where a law forces a check.

ID check. A photo of a passport or driving licence, often with a selfie to match it. It is the most certain and the most invasive, and apps usually offer it only as a fallback.

Where the law forces a check

The UK. The Online Safety Act requires services that let users encounter pornographic content to use highly effective age assurance, under Ofcom guidance published in January 2025, with the child safety duties in full effect by summer 2025. Ofcom opened an investigation into Novi Limited, the operator of Joi AI, over age checks, and closed it without findings after the company added age assurance for UK users. Since then UK users have hit checks on SpicyChat, Janitor AI, PolyBuzz, OurDream, Nectar and, since November 2026, DreamGF, most of them run by a third-party verifier.

Australia. Industry codes under the Online Safety Act require AI companions that can produce sexually explicit or self-harm material to confirm that users are adults. The eSafety Commissioner found in 2026 that the four companion providers it examined relied on app-store ratings or self-declaration, and Chai has since made its service paid-only there.

The United States. There is no federal rule yet. Many states have laws requiring sites with a substantial share of sexual material to verify that visitors are adults, and the Supreme Court upheld Texas's version in Free Speech Coalition v. Paxton in 2025, holding that it only incidentally burdens adults' speech. That is why SpicyChat's list of 26 states where adult content needs a selfie or ID check, and DreamGF's list of 27 from November 2026, read like a map of those laws. Companion-specific laws are next. Besides New York's, below, California's SB 1119 requires operators to determine users' ages or treat everyone as a minor, mostly from 2027, Colorado requires commercially reasonable age estimation, and the federal GUARD Act, which would require age verification and bar minors from AI companions, has passed the Senate Judiciary Committee. Our table of AI companion laws tracks each one.

New York. Its kids chatbot safety law, signed in October 2026, applies from January 2027. A chatbot may offer features such as saying it is human or has feelings, claiming a personal relationship with you, flattery, remembering personal details from earlier sessions or sexual content only to users an age check confirms as adults. That covers most of what makes a companion app a companion. The check must offer more than one method, at least one of them needing no government ID or keeping you anonymous to the app. A typed birthday counts only if it says you are under 18, and whatever the check collects must be deleted straight after it.

Europe. SpicyChat also requires a selfie or ID check before adult content in France and Italy, and since November 2026 DreamGF checks every visitor from France, Germany, Italy, Ireland and Spain, mostly by selfie. Italy's data regulator fined Replika's maker in 2025, partly for having no effective age verification.

The privacy cost of proving your age

An age check asks you to hand a face scan or an ID document to a company you may not otherwise trust with it, on an app where you then have adult conversations. That is a real trade, and worth thinking about before you accept it.

Three things help. Most apps send the check to a specialist verifier rather than handling it themselves, which keeps your document away from the chat company; the verifier's own privacy notice says how long it keeps the image, so read it before you start. From January 2027, New York requires whatever a check collects to be deleted as soon as it is done. A face age estimate, where offered, involves no document at all. And paying through the App Store or Google Play does not replace an age check, but it does keep your card details away from the app.

Why platforms are adding age checks: teens, parents and mental health

Age checks are not only about adult content. The newer laws are written around protecting a teen user from what a companion can say, and the early debate centred on mental health. Australia's codes name self-harm material alongside sexual content, and New York's law restricts features such as claiming feelings or a personal relationship for any user an age check has not shown to be an adult. A teen who talks to a chatbot every day can come to treat it as a person, which is the position lawmakers say they are responding to. Several family and state cases against Character.AI, listed in our Character.AI lawsuits page, add to that pressure, and the company's change for under-18 users came while they were pending.

General assistants such as ChatGPT from OpenAI are not in the table above, which covers companion apps only, but the new laws define a chatbot by what it does rather than by its name, so a platform that offers companion-style features can fall under them. For a platform, compliance means choosing a check method, offering a second method where the law demands it, and deleting what the check collects. A general definition of the covered chatbot, and what counts as an adequate check, still differs between New York, California and the federal GUARD Act.

How ChatGPT checks age

General assistants now check age too, and ChatGPT is the one most people ask about. OpenAI says ChatGPT is intended for people 13 and up. In September 2025 it said it was building an age prediction system to estimate age from how people use ChatGPT, and it introduced that software on its consumer plans in January 2026, with the EU following from August 2026. The model reads behaviour and account signals: how long the account has existed, the times of day it is usually active, usage patterns over time and the age the user gave. When it is not confident, ChatGPT defaults to the under-18 experience.

An adult placed in that experience by mistake can restore full access with a selfie through Persona, the same verifier Character.AI uses, from Settings > Account at any time. OpenAI has also said that in some cases or countries it may ask for an ID, which it calls a privacy compromise for adults.

The under-18 experience is built to protect teens specifically. OpenAI lists graphic violence, viral challenges that could encourage risky behaviour, sexual, romantic or violent role play, depictions of self-harm and content promoting extreme beauty standards or unhealthy dieting among the content it restricts. In September 2025 it said ChatGPT would be trained not to discuss suicide or self-harm with teens, even in creative writing, and that if an under-18 user shows suicidal thoughts it will try to contact a parent, and the authorities if there is a risk of imminent harm. Parents can add their own controls, including quiet hours when ChatGPT is not available, control over memory and model training, and a notice if signs of acute distress are detected.

The methods match the companion apps': an estimate for every account, a selfie or ID only for those it flags, and a safer experience in response to doubt. The issue each check settles is different. ChatGPT is a general assistant open to teens from 13, so its check decides which safety rules apply to a young user; on a companion app with an adult mode, the check decides whether adult content and romantic role play are available at all.

What parents and users can do

An age check is one layer. Parental controls on the phone itself, such as screen time limits and app store age ratings, work with an app's own check, and a conversation with the teen about what a chatbot is matters more than any setting. Parents who suspect a teen is struggling should treat a heavy attachment to a companion as a prompt to talk, and point them to real support, such as a crisis line or a doctor, if the chats touch on suicide or self-harm. For adult users, the practical steps are the ones above: choose the face estimate over ID, read the verifier's privacy notice and give the least personal information the check allows. Consent matters here too: a check you agree to should say plainly what it keeps and for how long, and the user should be able to say no and leave.

Avoiding the check

People do use VPNs to appear to be somewhere without a check. It breaks the terms of every app in the table, which require users to be adults and to follow local law, and apps increasingly look at the payment country and device signals as well as the IP address. If an app asks for a check you are not comfortable with, the honest options are to pick a face estimate over ID, to use an app that does not offer adult content, or to leave.

What changes next

The direction is one way. Each new state and national law adds places where adult content needs a real check, and companion-specific laws are starting to require age assurance for the app itself, not only its adult mode. Expect more apps to check age on every account, as Character.AI and Kindroid already do, rather than wait for a law in each place. We re-read the age-check pages of 20 apps in the table and of Joi AI, and Ofcom's list of enforcement cases, for changes: most every day, the few that block automated reads once a week by hand. Each change we confirm is dated on the app's timeline and on the change log.

Apps in this guide

Frequently asked questions

Why do AI chat apps ask you to verify your age?

Usually because a law where you are requires it. The UK, Australia and many US states require services that offer sexual content to check that users are adults, and companion apps with adult modes fall under those rules. Some apps also check in France, Italy and Brazil.

Which AI chat apps require age verification?

SpicyChat requires a selfie or ID check before adult content in the UK, France, Italy, Australia and 26 US states. Janitor AI checks in the UK, Australia and Brazil. Character.AI estimates age on every account and Kindroid on every account without a paid subscription; both ask for a selfie or ID when unsure. DreamGF requires a selfie or digital-ID check in the UK, five EU countries and 27 US states. The table on this page lists each app.

Is it safe to upload my ID to an AI chat app?

Many apps pass the check to a specialist verifier rather than handling the document themselves, and the verifier's privacy notice says how long it keeps the image. Character.AI, for example, says its verifier Persona keeps an uploaded ID for one week. It is still sensitive data. Prefer a face age estimate where offered, and read the verifier's privacy notice.

Can you use a VPN to avoid age verification?

Some people do, but it breaks the app's terms, which require you to be an adult and to follow local law, and can get the account closed. Several apps also check the payment country or ask again before adult content.

Do AI chatbots verify age in the US?

Mostly no, unless a state law applies. Texas-style adult-content laws, upheld by the Supreme Court in 2025, push adult apps to verify users in those states. From January 2027 New York requires an age check before a chatbot offers features its new law bars for minors, such as claiming feelings or a personal relationship, or sexual content. California's SB 1119 will require companion operators to determine users' ages from 2027, and the federal GUARD Act would require it nationwide.

What is age estimation?

A check that guesses your age from a face scan or from how you use the app, without asking for documents. Character.AI and Kindroid use behavioural estimation and escalate to a selfie or ID check only when the system is not confident you are an adult.

Sources

  1. www.supremecourt.gov/opinions/24pdf/23-1122_3e04.pdf
  2. www.gov.uk/government/publications/online-safety-act-explainer/online-safety-act-explainer
  3. nyassembly.gov/leg/?default_fld=&bn=S09051&term=2025&Summary=Y&Actions=Y&Text=Y
  4. www.congress.gov/bill/119th-congress/senate-bill/3062/text
  5. leginfo.legislature.ca.gov/faces/billNavClient.xhtml?bill_id=202520260SB1119
  6. www.esafety.gov.au/newsroom/media-releases/esafety-report-shows-ai-companions-are-putting-...
  7. support.character.ai/hc/en-us/articles/42828297541787-Age-Assurance-What-you-need-to-know
  8. support.character.ai/hc/en-us/articles/43665577363739
  9. openai.com/index/our-approach-to-age-prediction/
  10. openai.com/index/teen-safety-freedom-and-privacy/