AI companion age verification: how services check that you are 18

Services that allow adult themes have to keep minors out, and laws in several countries now require more than a checkbox. For users, the questions are practical: what will I be asked for, who sees my face or ID, and what does the service keep afterwards?

This guide compares what companion services publish about age checks on their own pages, checked on October 23, 2026, and summarizes the legal background. It is general information, not legal advice. We do not explain how to get around any age check.

• Self-declaration. You click "I am 18 or older" or enter a birth date. Fast and private, but easy to lie on, so it offers little protection.
• Payment as a signal. Some services treat a paid subscription as evidence of adulthood, because card holders are usually adults.
• Facial age estimation. You take a live selfie and software estimates whether you look over a threshold. Liveness checks try to stop people from holding up a photo.
• ID document check. You photograph a government ID, often with a selfie to match it.
• Reusable age tokens. Some providers let you prove age once and reuse the result elsewhere, so you do not repeat the check for every site.

The stricter methods are usually run by a specialist provider, not by the companion service itself. Well-designed setups send the service only a yes or no answer.

Service	Age check described on official pages
Candy AI	Age gate where users affirm they are over 18; the Underage Policy says the operator "may also implement further measures" where local rules require
GirlfriendGPT	Users affirm they are over 18; no document-based check is described for general access
PolyBuzz	Date of birth on first use; in the UK, mandatory verification through Ondato or Yoti; PolyBuzz keeps only "verified" or "not verified"
Janitor AI	In Australia, Brazil and the UK, a one-time check through k-ID by facial age estimation, ID scan or an existing AgeKey; Janitor receives only a yes or no
OurDream	Age confirmed at sign-up; where law requires (the UK is named), verification through Go.cam or AgeVerif, with only a pass or fail returned
Secrets AI	Age gate, checks at registration and verification on request
DreamGF	UK access restricted until a supported verification process is available
Nectar AI	NSFW mode limited to verified users 18+; the method is not described, and its terms, FAQ and privacy policy disagree on age
Kindroid (kindroid.ai)	Paid subscribers are treated as verified; free users can be flagged and asked to verify through Yoti, by selfie estimation or ID
Character.AI	Announced in-house age assurance combined with Persona tools; since November 25, 2025, users under 18 cannot have open-ended chats

The pattern: most services rely on self-declaration in most of the world and switch to selfie or ID checks only where a law forces them, most often in the UK. Sources for each row are listed at the end.

When a service uses an outside provider, two privacy policies apply: the service's and the provider's. Before you upload anything, check:

1. Who runs the check. The service should name the provider. PolyBuzz, for example, names Ondato and Yoti and links to their privacy pages.
2. What the service receives. The better answer is a yes or no result only. Janitor AI, OurDream and PolyBuzz all say this is what they get.
3. What the provider keeps. Janitor's help article says k-ID does not keep the document or face data after the check. Look for a similar statement from whichever provider you meet.
4. Whether estimation is offered. A selfie estimate reveals less than an ID document, which carries your name, address and document number.

Never send an ID by email or chat to someone claiming to be support. Legitimate checks run inside an official flow.

Around half of US states have passed laws requiring age verification for websites with a significant share of sexual material that is "harmful to minors"; most use a threshold of one third of the site's content, and counts differ between trackers. In Free Speech Coalition v. Paxton (June 27, 2025), the Supreme Court upheld Texas's law in a 6 to 3 decision (opinion).

How these laws apply to AI chat services, where content is generated in private conversations rather than published, has not been settled. That is one reason US users often see only a self-declaration gate while UK users see selfie or ID checks.

Since July 25, 2025, services in scope of the UK Online Safety Act that allow pornographic content must use "highly effective age assurance", a standard that a simple checkbox does not meet. This is why PolyBuzz, Janitor AI, OurDream and others describe specific UK checks, and why DreamGF restricts UK access until it supports one. The PolyBuzz privacy policy names the Online Safety Act as the legal basis for its UK checks.

Two US states regulate companion chatbots directly. We use their requirements as review criteria, so it helps to know what they say.

New York, General Business Law Article 47 (in force since November 5, 2025): operators of AI companions must tell users clearly that they are not talking to a human, at the start of a session and every three hours of continued use, and must have a protocol to detect suicidal ideation or self-harm and refer users to crisis services. The state Attorney General enforces it, with civil penalties of up to $15,000 per day.

California SB 243 (in force since January 1, 2026): operators must give a clear notice that the chatbot is AI where a reasonable person could be misled, keep a protocol that prevents content about suicide and self-harm and refers users to crisis providers, and publish details of it. For users the operator knows are minors, the law adds AI reminders at least every three hours and measures to stop sexually explicit material. Annual reports to the state's Office of Suicide Prevention start in July 2027. Injured users can sue for at least $1,000 per violation.

Neither law creates an age verification requirement on its own, but SB 243's rules for known minors push operators to find out who their users are.

• Expect a self-declaration gate in most of the US and a selfie or ID check in the UK and some other countries.
• Prefer services that name their verification provider and receive only a yes or no.
• Prefer facial estimation over uploading an ID where you have the choice.
• Do not use VPNs or borrowed documents to avoid a check. It breaks the terms of every service listed here, and Janitor, for example, detects the requirement by IP address.
• Treat a service with no visible age rules as a red flag, especially if it allows adult content.

For what services do with the rest of your data, see the AI companion privacy guide. Our methodology explains how we use these rules when describing services.

• Candy AI Underage Policy
• GirlfriendGPT Underage Policy
• PolyBuzz Privacy Policy (updated June 1, 2026) and Terms of Use
• Janitor AI: age verification for Australia, Brazil and the UK
• OurDream Trust and Safety: age verification
• Secrets AI Underage Policy
• DreamGF: UK users
• Nectar AI FAQ and terms
• Kindroid age verification
• The Record on Character.AI age assurance (October 29, 2025)
• Free Speech Coalition v. Paxton, Supreme Court opinion
• California SB 243 bill text
• New York General Business Law Article 47

Checked October 23, 2026. Laws and policies change; check the current text before relying on it.