Why is ChatGPT asking for my ID?
ChatGPT asks for ID when its age prediction model has guessed that your account belongs to someone under 18 and moved you into ChatGPT for Teens, or, in Italy, because a law makes every user prove their age. The first case is the common one. OpenAI switched age prediction on for consumer plans on 20 January 2026, extended it to the EU on 25 August 2026, and it runs whether or not you gave a birthday at signup. OpenAI's help page lists the signals: "general topics you talk about, the times of day you use ChatGPT, how and when your account is used, and how long your account has existed." When the model is unsure, the January announcement says, "we default to a safer experience."
The ID request is the exit, not the trigger. Nothing you typed makes ChatGPT demand a document; the model classifies the account, the account gets teen protections, and the Verify age button in Settings is offered as the way out. TechRadar, reporting on 29 January 2026, nine days after launch, described the adults it had caught: newer accounts, late-night users and people asking about teen-relevant topics, some of them long-standing Pro subscribers. Sam Altman had said this would happen. His 16 September 2025 post announcing the system reads: "In some cases or countries we may also ask for an ID; we know this is a privacy compromise for adults but believe it is a worthy tradeoff."
Italy is the other case, and it is a legal requirement rather than a guess. OpenAI's age verification page says "users in some locations are required to verify their age before continuing to use the service", names Italy, and warns that an account which misses the deadline in its notice is deactivated until the check is done. If your screen shows a deadline, you are in this case, and the rest of this page about refusing does not apply to you.
What is restricted until you verify?
Until you verify, ChatGPT keeps the ChatGPT for Teens protections on your account and changes nothing else: the account stays active, your plan keeps running, and, in OpenAI's words, "OpenAI will not show ads in your account." The help page is explicit that verification is not a condition of use: "Age prediction does not require you to verify your age to keep using ChatGPT with teen protections."
What the teen experience reduces, from the same page as of September 2026: "Graphic violence or gore", "Viral challenges that could push risky or harmful behavior", "Sexual, romantic, or violent roleplay", and "Content that promotes extreme beauty standards, unhealthy dieting, or body shaming". The January 2026 announcement adds "Depictions of self-harm". Code, drafting, study, files, search and voice are not on the list. A parent can link the account through parental controls, and OpenAI says "Parental controls do not give your parent or guardian access to your conversations."
Adult mode was the other half of this trade, and it never arrived. OpenAI said in October 2025 that verified adults would get erotica in December, moved that to the first quarter of 2026, then told TechCrunch on 7 March 2026 it was "pushing out the launch of adult mode" to work on higher priorities, with no new date. As of September 2026 the age check removes teen filters and adds nothing an adult account did not already have.
What happens to your ID after ChatGPT age verification?
After ChatGPT age verification, Persona deletes your selfie or ID within 7 days and OpenAI never holds the image at all. The sentence is OpenAI's: "Persona handles age verification and deletes your uploaded ID or selfie within 7 days after verification. OpenAI does not receive your ID or selfie. OpenAI only receives your date of birth or an age prediction."
What Persona asks for depends on the country. OpenAI says it "may ask for one or both" of a live selfie, taken on your phone or webcam so it cannot be a photo of a photo, and a government ID: passport, driver's licence, state or provincial ID, national ID card, residency permit or employment authorization card. The January 2026 announcement describes the selfie as the normal route ("restore their full access with a selfie through Persona"), so the document is what comes out when the selfie is not enough, and in Italy, where the help page says Persona may ask for "a live selfie, a government-issued ID, or both". If you cannot take a selfie at all, the ID stands in for it.
The detail most people miss is on the FAQ: "You do not need to show ID numbers, such as a national ID number or Social Security number. You may cover those before uploading." The photo has to show your full name, date of birth and face, and nothing else. Two other lines are worth acting on. First, verifying ends the guessing: "After you verify, OpenAI will not run age prediction on your account." Second, Italy's recovery path for an already deactivated account runs through Stripe rather than Persona, and OpenAI's page on that path says validation "usually takes a few hours" and "Your ID document data is deleted immediately after the age verification process is complete."
What to do if ChatGPT wrongly flagged you as under 18
Verify once, on the web, and cover your ID number if a document is asked for. The whole path is in Settings, and the first step is deciding which of the two cases you are in.
- Read the notice. A required age-verification notice with a deadline is the legal case (Italy); a teen-mode account with no deadline is age prediction, and you may ignore it if the filters do not touch your work.
- Sign in on the web, open Settings, then Account, then Verify age, or go to chatgpt.com/verify_age while logged in. The mobile apps send you to the same Persona flow.
- Take the live selfie when it is offered. Persona checks your face against the age it estimates, and for most adults that is the end of it.
- If a document is requested, use a passport or licence that shows your name, date of birth and photo, and cover the ID number before you photograph it.
- Wait. OpenAI says the change "may take a short time to fully apply". If protections stay, the check did not clear you, and OpenAI says it "may keep or restore teen protections".
- Read the error. "Unfortunately, your age could not be verified. Please try again later" means you cancelled midway; start again. "Please contact support" means you have hit the limit of unsuccessful attempts, and only OpenAI Support can reset it.
- To object to the prediction itself rather than pass it, OpenAI's address is [email protected]; the help page invites objections there.
Two edge cases from OpenAI's page on turning 18. A real teen who is told they now "have access to more features & settings" can pick "I'm under 18" to run the same verification and keep the protections. And a parent cannot do it for them: "Parents and guardians cannot verify a teen's age on their behalf. Only the teen can complete the age verification process for their own account." An account that verifies as under 18 stays in ChatGPT for Teens until the verified date of birth says otherwise, and OpenAI now notifies both sides about a week before that switch.
Do Gemini, Claude, Copilot, Perplexity, Grok and Meta AI ask for ID too?
Four of the other seven can ask an adult for a selfie or ID, and none of the eight asks at signup. We compiled each assistant's minimum age and age-check rule from its own terms and help pages on 27 September 2026. The table is for personal consumer accounts; work and school accounts have their own rules.
| Assistant | Minimum age | Can it ask an adult for a selfie or ID? | What triggers it | Where the image goes |
|---|---|---|---|---|
| ChatGPT | 13; under 18 with a parent's permission (Terms, 1 January 2026) | Yes | Age prediction places the account in ChatGPT for Teens, or a required notice in Italy | Persona; deleted within 7 days; OpenAI never receives it |
| Gemini | 13, or your country's age, on a personal Google Account; 18 on a work account | Yes, through the Google Account | Google's age estimation (US accounts since 30 July 2025) or an age-restricted feature | Google: a government ID "is deleted after your date of birth is successfully verified"; a credit card or digital ID also counts |
| Claude | 18 | No ID route exists | The age on the Apple or Google account; Anthropic sends disputes to App Store or Play support | Nothing reaches Anthropic |
| Copilot | 13, higher in some countries (Microsoft) | None described on the Copilot age page | The Microsoft account's birthdate | Not applicable |
| Perplexity | 13; a parent must accept the terms for a minor (Terms, 23 January 2026) | None described | Self-declared | Not applicable |
| Grok | 13; 13 to 17 with a parent's permission (xAI consumer terms, 11 September 2026) | Yes, through an X account | X's own age check: a live selfie estimated by xAI's model, or a government ID, added for the UK's Online Safety Act | X; no retention period published |
| Meta AI | 13, or your country's age (Meta AI terms, 13 May 2026) | Yes, through the Instagram or Facebook account | Meta's AI age assurance moves the account into teen settings, or a birthday edit from under 18 to over 18 | Yoti estimates age from a video selfie and "Meta and Yoti then delete the image"; an uploaded ID "is deleted within 30 days" |
| Whizi | Not directed to children under 13 (privacy policy, 13 August 2026); minors need a parent's permission (terms) | No age check described; identity is verified only for a privacy-rights request | Nothing; there is no age prediction and no teen mode | Not applicable |
Two patterns fall out of the table. The assistants that can ask for a document are the ones bolted to a large consumer account (Google, X, Meta), plus ChatGPT, which built its own system. The ones that cannot are the ones that handed age to the app store or to the terms. Claude has the strictest floor and the lightest check at once: Anthropic requires 18, never sees a document because Apple or Google already holds your birthday, and its help page sends anyone who disputes the age to the app store rather than to a verification vendor. Google's route is the widest: its age-restricted content page accepts a government ID, a credit card or, on Android in some countries, a digital ID, and 9to5Google reported on 30 July 2025 that a selfie option is processed on the device.
Whizi, which publishes this page, sits in the terms-only row. Its privacy policy says the service is not directed to children under 13, its terms say minors must have a parent's permission and supervision, and the only identity check either document describes is one Whizi "may need" before completing a privacy-rights request. There is no age prediction, so an adult using GPT, Claude or Gemini through it is never asked for a selfie. The flip side is real: Whizi has no teen mode and no parental controls, so a household that wants those should stay with ChatGPT or Gemini. What Whizi does with your data covers the rest of the policy.
Verify, keep the teen version, or switch: the decision rule
Verify if the teen filters touch your work (fiction with violence or romance, body and diet questions, dark subject matter) and you can accept Persona holding a selfie for up to 7 days. Keep the teen version if your use is code, study, drafting and search; every one of those survives, and you hand over nothing. Switch if you will not give a document to anyone and the filters block you: Claude asks nothing beyond the age on your app store account, Copilot and Perplexity check nothing beyond a birthday, and AI without a phone number and using AI without an account list what each one asks at the door.
Whichever you pick, read the notice before you decide. A deadline means Italy's legal check, not the prediction, and letting it lapse deactivates the account rather than filtering it. Everywhere else, the worst outcome of doing nothing is a chatbot that declines to write gore.
- Read the notice first: a deadline means a legal requirement (Italy); no deadline means age prediction, and you can ignore it
- Open Settings, Account, Verify age on the web, or chatgpt.com/verify_age, and let Persona try the selfie before any document
- If a document is asked for, cover the ID number; OpenAI says only your name, date of birth and photo need to be visible
- Expect the image to sit at Persona for up to 7 days and never reach OpenAI; verifying once also stops age prediction on your account
- Stuck on "Please contact support"? You hit the attempt limit; write to OpenAI Support, or to [email protected] to object to the prediction itself
- Refusing costs you the teen filters, not the account, unless the notice was Italy's required check
Frequently asked questions
Does ChatGPT ask for ID?
Yes, in two situations. If OpenAI's age prediction places your account in ChatGPT for Teens, you can verify with a selfie or government ID through Persona to remove the protections, and that is optional. In Italy, a legal requirement means every user gets a required age-verification notice, and an account that misses its deadline is deactivated until the check is done.
Is it safe to give ChatGPT my ID?
The ID never reaches OpenAI. Persona runs the check, deletes the uploaded ID or selfie within 7 days, and passes OpenAI only your date of birth or an age result. OpenAI also says you may cover the ID number before uploading, because only your name, date of birth and photo need to be visible. Whether a week at a verification vendor is acceptable is your call; the alternative is keeping the teen filters.
Can I use ChatGPT without verifying my age?
Yes, outside Italy. OpenAI says "Age prediction does not require you to verify your age to keep using ChatGPT with teen protections." Your account, plan and history stay as they were; the only change is reduced sensitive content such as gore, sexual or violent roleplay and self-harm depictions, plus no ads. In Italy the check is required by law and the account is deactivated if you skip it.
How old do you have to be to use ChatGPT?
13, and under 18 needs a parent or guardian's permission, under OpenAI's terms of use effective 1 January 2026. Claude requires 18. Gemini, Copilot, Perplexity, Grok and Meta AI all set 13, or a higher age where a country requires one, and Whizi's policy says its service is not directed to children under 13.
