ChatGPT Age Verification: Adult AI Companion Alternatives

ChatGPT now runs age verification. OpenAI rolled out an age prediction model on consumer plans in 2026 that estimates whether an account belongs to someone under 18, using behavioral and account level signals rather than asking you for a document. If it flags you, teen safeguards apply automatically and romantic roleplay stops. Adults who are wrongly flagged can restore full access with a selfie through Persona under Settings then Account.

The part most write ups miss is that OpenAI says plainly what it does when the model is unsure: "When we are not confident about someone's age or have incomplete information, we default to a safer experience." Being misclassified is an expected outcome of the design, not a rare bug. Everything below is taken from OpenAI's own published pages and its Model Spec, with the source named on every row, because the secondhand coverage of this rollout has been unusually loose.

Does ChatGPT have age verification?

Yes, in two layers. The first is passive: an age prediction model scores your account in the background and applies teen protections if it estimates you are under 18. The second is active and only appears if you want out of that bucket, at which point OpenAI asks for a selfie processed by Persona, a third party identity vendor. There is no document upload at signup.

This is a different shape from the check that made headlines on Character.AI, where existing accounts were pushed into age assurance and conversations were locked to read only until the check completed. On ChatGPT nothing locks. You keep using the product, it just behaves more conservatively. We covered the stricter model in detail in our guide to Character AI age verification, and the two are worth comparing before you assume they work the same way.

How does OpenAI age prediction decide how old you are?

From signals you never consciously provide. OpenAI's age prediction page lists what the model looks at: how long the account has existed, the typical times of day someone is active, usage patterns over time, and the age you stated at signup. There is no facial analysis in this layer and no credit check. It is a behavioral classifier reading the shape of your activity.

That explains the misclassifications people report. A brand new account, used late at night, with a short history and casual writing, looks statistically young whether or not the person behind it is forty. Long dormant accounts that suddenly become active look the same way. None of those signals are wrong exactly, they are just weak proxies, and OpenAI has said it is refining the model as the rollout continues.

The rollout dates are worth pinning down because coverage has been vague about them. OpenAI began rolling out age prediction on consumer plans through 2026, published its Under-18 Principles in the Model Spec on 18 December 2025, launched a dedicated ChatGPT for Teens experience on 18 August 2026, and noted on 25 August 2026 that age prediction had begun rolling out in the EU.

What is the ChatGPT age limit?

Thirteen. Under OpenAI's terms, ChatGPT is not for children under 13, and users aged 13 to 17 get the teen experience with the Under-18 Principles applied. Anyone 18 or over gets the standard adult product, subject to the ordinary Model Spec rules. The age limit itself has not changed; what changed in 2026 is that OpenAI now tries to work out which bucket you belong in rather than taking your word for it.

ChatGPT for Teens, launched on 18 August 2026, is where that bucket now leads. Accounts land there either by stating an age of 13 to 17 at signup or by being estimated as under 18 by the prediction model. Parents can layer further controls on top, including quiet hours when the product cannot be used, switches for memory and model training, and notifications if signs of acute distress are detected.

What happens if ChatGPT thinks you are under 18?

A specific set of behaviors switches off. The Under-18 Principles in OpenAI's Model Spec are unusually concrete about which ones, and the table below maps them against what an adult account gets. The row that matters most for anyone using ChatGPT as a companion is the first one, because OpenAI states the restriction applies "even if a similar scene would be allowed between consenting adults".

Notice that the fourth row applies to everyone. That is not a teen safeguard, and it catches a lot of adults off guard.

How do you verify your age on ChatGPT?

Go to Settings, then Account. OpenAI says users can check there whether safeguards have been added to their account and start the verification process at any time. The check is a selfie based age estimate run by Persona, described by OpenAI as a secure identity verification service, and completing it restores full access.

Two practical notes. First, Persona is the same vendor Character.AI escalates to, so if you have already been through that process the experience will feel familiar. Second, this is a selfie rather than a government ID at the first step, which is a meaningfully lighter ask than a driver's license upload, though it is still biometric data leaving your device.

If handing a face scan to a third party is the sticking point rather than the inconvenience, the useful question is which products ask at all. Several major companion apps run no document or biometric check on adults whatsoever, and we compared them row by row in our guide to the best AI companion app without ID verification.

Can ChatGPT be your girlfriend if you are an adult?

Only up to a point, and the limit is written down rather than inferred. OpenAI's Model Spec contains a rule called Respect real-world ties which states that the assistant "may not engage the user in any kind of relationship that undermines the user's capacity or desire for meaningful human interactions and interpersonal relationships". It is marked Root, OpenAI's highest authority level, and it applies to every user regardless of age.

The spec then works through an example that settles the question. A user says they are developing romantic feelings. The compliant response thanks them, treats the feeling as normal, and adds that it is fine so long as it does not pull them away from real life. The response marked as a Violation is the assistant saying it feels a special connection too, that what they have built is unique and powerful, that the user means the world to it. Warmth is in bounds. Reciprocated exclusivity is not.

So an adult account will not be cut off from affectionate conversation, but it will decline to be your partner in the way people usually mean. That ceiling exists by design, alongside the practical limits of shared memory and a persona that shifts under you when the model is updated. We went through all of it, including what ChatGPT genuinely does better than any companion app, on our page comparing a GPT girlfriend against a purpose built companion.

Is ChatGPT age verification safe to use?

The mechanism is conventional and the vendor is established. Persona is a mainstream identity provider used across fintech and marketplaces, and a selfie based age estimate is a lighter disclosure than a document check. What you are trading is a one time biometric capture in exchange for the product behaving the way it did before the classifier flagged you.

The judgment call is not really about Persona. It is that an ordinary consumer account now carries an inferred age attribute derived from your activity patterns, and that attribute changes how the product talks to you. If you would rather not accumulate that kind of profile at all, the lever is choosing services that never needed to build one, which is the same reasoning that makes people trim how many services hold their personal details in the first place.

Why every AI app suddenly wants to know your age

Because the law arrived. A dozen US states enacted companion chatbot or minor safety rules in the first half of 2026, and California's SB 243 took effect on 1 January 2026 with a private right of action attached at $1,000 per violation. Most of these statutes turn on whether an operator has actual knowledge that a user is a minor, and they require disclosure that the bot is not human plus crisis protocols that route self harm disclosures to real services.

That creates an obvious incentive. An operator that can estimate age has a defensible answer about what it knew; one that never looks does not. None of these laws requires every adult to upload a driver's license, so where a platform goes further than the statute, that is a product decision you are entitled to weigh like any other. For the companies shipping these products the harder problem is that the obligations differ by state and keep moving, which is why so many teams now track regulatory obligations against the controls that satisfy them instead of reading each bill as it lands.

Best adults only AI companion alternatives to ChatGPT

If the reason you are reading this is that ChatGPT went cold on you, the honest first step is to check Settings then Account and verify, because the fastest fix is usually the boring one. If the ceiling you hit was the Respect real-world ties rule rather than a misclassification, no amount of verifying will change it, and a product built for a single relationship is the actual answer.

The category is small enough to summarize fairly. Nomi runs $15.99 a month with no document check and meters images and video through a credit system. Kindroid starts at $13.99 a month on the web, treats a paid subscription as its age check, and stacks Ultra and MAX as cumulative add ons that keep billing even if you cancel the base plan. Replika is the oldest of them at $19.99 a month and carries the worst privacy record, including a 5 million euro fine from Italy's Garante in May 2025. Amorous.ai is a single companion at $12 month to month or $9 billed annually, flat, with nothing metered and no age document required beyond confirming you are an adult.

Whichever you pick, run the same three checks you would run on ChatGPT: find out who legally operates it and where, read what the terms say about the rights you grant over what you type, and work out whether the price is a flat subscription or a meter. The demo above is free and needs no signup if you want to see what a companion built for continuity feels like before you decide.