跳到主要内容
Didit 融资 750 万美元,打造身份与欺诈基础设施
Didit
返回博客
博客 · 2026年11月22日

Age assurance for apps: games, social, dating, AI and more

One decision table for product owners: what each app category has to do about age in the US, Brazil, Australia and the EU, when the store signal is enough, which feature to gate and which check to add.

作者:Didit更新于
age-assurance-mobile-apps-by-category-cover.png

In short

Age assurance for an app depends on the feature, not on the app store category. A utility app can usually stop at the store's age signal, while betting, paid loot boxes and sexual content in Brazil need verification beyond it.[2][4]

  • Where the signal answers and no law or risk asks for more, stop there.
  • Gate the risky feature; gate the app only when all of it is for adults.

Last reviewed: 22 November 2026 · Not legal advice

Age assurance is the umbrella term for verifying, estimating or inferring a user's age or age range. Verification is its stricter part: in Brazil, a procedure with a high degree of reliability.[2]

Google says its Play Age Signals API will return responses for all users by 11 December 2026, and that the user's sharing becomes mandatory in California and Brazil from January 2027.[24][23] Google does not require developers to use it: see our Play Age Signals guide.[26] This page covers what your kind of app does next.

Not sure which row is yours

Bring your features and markets. Get a demo of age estimation in your app, or start free.

How to read the table

Run these steps once per feature.

1List the laws that make you receive the signal

California, Brazil, and the store laws of Texas, Alabama, Utah and Louisiana, most from 2027.[5][1][12][10][11][13]

2Mark features restricted by law or by your terms

Random boxes, feeds, alcohol, bets, sexual content, minimum ages.

The signal answers and no law or risk asks for more

Yes

Stop at the signal

Apply the age-appropriate experience. Collect nothing else.

No

Restrict by default, then offer a check

No signal keeps a restricted feature closed. Check only the people who ask for it.

Confirm your laws with counsel.

From 1 January 2027 a developer in California's scope "shall request a signal" when the app is downloaded and launched, whatever its audience, and must not prompt the user to change their age information.[5] See our AB 1043 and AB 1856 guide, and the App Store Accountability Acts guide for the other states, dates and court cases. In Brazil the law asks services aimed at minors, or likely to be accessed by them, to ensure they receive the signal, and receiving it does not end their responsibility; the decree's receipt rule covers improper, inadequate or prohibited offers, so scope is for counsel.[1][2] See our Brazil guide.

Age assurance by app category: the decision table

"Draft" is the preliminary guide of ANPD, Brazil's data protection authority; we found no final version on 22 November 2026.[4] A market missing from a row means we found no trigger in our sources, not that none exists. Methods are design choices, not words of the law.

App categoryLegal trigger by marketIs the store signal enoughFeature to gateMethod to add
Game, all agesBrazil: not every game has to run age assurance[3]Usually yes. Draft: low risk, signal or verifiable credentials[4]NoneNone
Game with chat or microtransactionsBrazil: interaction limited by default[1]. Texas: each in-app purchase carries an age rating[12]Often, with protective defaults. Draft: moderate risk[4]Chat with strangersSelfie estimate if your risk assessment asks for it
Game with loot boxes, or rated R18+Brazil: verify, or keep the boxes off by default[2]. Australia: age checks for R18+ online games[17]Brazil: no, if the boxes stay reachable. Draft: verify regardless of the signal received[4]The paid random boxOff by default; document, liveness and face match to open it
Social app or feedCalifornia: no addictive feature under 16, from 2027 on our reading[6]. Brazil: accounts up to 16 linked to a guardian[1]. Australia: no accounts under 16[29]California, AB 1709: yes when it answers[6]. Australia: a bare self-declared age is not[19]Profiled feed, autoplay; the account in AustraliaSelfie estimate with a band, document inside the band
DatingApple: 18+ rating[27]. Brazil: apps mainly for sexual encounters are prohibited for minors[2]Brazil, sexual purpose: no. Elsewhere: a risk decisionThe accountSelfie estimate, document in the band; in Brazil, verification if the purpose is sexual
General AI assistantBrazil draft: moderate risk[4]. EU: proposal only[21]Often, with content controlsAdult outputControls; if adult output stays, the next row
AI companion with sexual or romantic contentApple: 18+ rating[27]. Brazil: sexually explicit AI dialogue equals pornography[2]. Australia: age checks if safeguards are missing[17]Brazil, explicit: no. Australia: not a bare self-declared age[18]The account or the explicit modeDocument in Brazil; estimate or document in Australia
Licensed streaming, music, booksBrazil: exempt with child profiles and parental controls[2]. California: autoplay may be in scope[6]Usually yesAdult profile; autoplay in CaliforniaNone for the catalogue
Marketplace or deliveryBrazil: verify for alcohol, tobacco, weapons, fireworks[2]General goods: usually. Prohibited goods in Brazil: noCheckout of those itemsDocument, liveness and face match
Betting and lotteriesBrazil: verify, ban and remove minors' accounts[2]. EU guidelines recommend verification[20]NoThe accountDocument, liveness and face match
EducationCalifornia: design code if likely accessed by children[7]. Brazil draft: low risk[4]Usually yesOpen community areasNone; child protections for everyone
Utility or productivityCalifornia: request the signal[5]. Brazil draft: low risk[4]YesNoneNone

Rules for every row

In Alabama, Utah and Brazil your own check can only tighten: the lower or more protective age wins for the settings these laws cover.[10][11][2] In California the signal is the primary indicator unless you hold "internal clear and convincing information", and whether an age estimate reaches that bar is untested.[5] Never send Google's or Apple's signal to an age check vendor, an ad tool or analytics.[25] Selfies in Texas, Illinois and Colorado need a biometric notice and release step.[37]

Games and loot boxes

Brazil has a game-specific rule. Law 15.211 bans loot boxes in games aimed at minors or likely to be accessed by them, by age rating (Art. 20).[1] The decree asks games with loot boxes to verify age, and drops the requirement when a version without boxes exists or the feature is fully restricted by default.[2] The official Q&A confirms it: the limit is not on access to the game.[3]

1Ship with random boxes off in Brazil

Everyone plays. Nobody is checked.

The player asks for paid random boxes

Yes

Verify 18 with a document

With liveness and face match. If the store signal says under 18, the box stays closed whatever the check says.

No

Keep playing with boxes off

No check and no data collected.

A selfie estimate may come first, but it does not open the box.

Declaring loot boxes in Apple's rating questionnaire sets the Brazil storefront rating to 18+.[28] Chat is a separate feature: Art. 21 limits interaction by default in games likely to be accessed by minors.[1] See age verification for mobile games.

Social apps and feeds

California's AB 1709 bars an addictive feature, meaning a profiled feed of user media or autoplay, for users under 16 on platforms where it is a significant part of the service. The platform first verifies age through the signal; if it cannot, it relies on its own determination under Health and Safety Code 27001, whose rules are not final, and whether that check sits at 16 or 18 is for counsel.[6][8][9] See our California 2027 guide.

1California: request the signal at first launch

Do not ask the user to change the age they declared.

What the signal says before the feed or autoplay starts

16 or over

Full feed

No second check under AB 1709. SB 976 is a separate law.

Under 16

Followed accounts only

No profiling, no autoplay. An account is still allowed.

No answer

Same limits, then your own check

Selfie estimate, document in the band, only if the user wants the feed.

California flow. No final SB 976 rule names a method.[9]

Since 10 December 2025 certain social platforms in Australia must keep under-16s from holding an account; the regulator does not consider self-declaration on its own reasonable, and lets providers consider app store signals as one input.[29][19] In Brazil, a network with prohibited content offers a version without it, with no verification needed, or verifies age.[2] In the UK, where the government plans to set out by the end of 2026 how its under-16 ban will work, Ofcom lists facial age estimation, with a challenge age and a second check, among methods capable of being highly effective at 16.[22] From 2028 Illinois operators may rely on the signal and need not verify independently.[15] See age verification for social media and our social apps page.

Dating apps

Apple now asks every app about "Dating and Relationships", with answers due by 12 January 2027, and a yes sets a minimum rating of 18+ (A18 in Brazil), so the account is the gate.[27] Brazil prohibits for minors the apps whose main purpose is sexual encounters, and those must ban and remove minors' accounts.[2] Whether a general dating app fits that wording is for counsel.

  1. Sign-upSignalUnder 18 in the signal: no account.
  2. Before profileSelfieEstimate with liveness, set to a band around 18.
  3. Inside the bandDocumentPortrait matched to the selfie.
  4. LaterRe-checkReports that suggest a minor reopen the check.

In Brazil, an app mainly for sexual encounters must verify age.

Play restricts users it identifies as minors from downloading 18+ apps in Brazil, Australia, the EEA, Singapore, Switzerland and the UK.[23] Apple says of 18+ apps in Brazil, Australia and Singapore that developers "may have separate obligations to independently confirm that their users are adults".[28] Elsewhere it is a risk decision: write down why. See our dating page.

AI companions and chatbots

General-purpose generative AI sits in the moderate band of Brazil's draft; sexually explicit AI dialogue is treated as pornography.[4][2] Apple's 18+ descriptor names "a virtual companion that engages a user in sexual or romantic content".[27] In Australia the codes' definition of an AI companion chatbot feature leaves out features whose controls make the risk of restricted material immaterial.[18]

Route A: take the content out

Controls instead of checks

  • Block restricted output for everyone
  • Minor defaults when the signal is missing
  • No age check for adults

Fits general assistants

Route B: keep the content in

Gate the mode at 18

  • Explicit mode off by default
  • Selfie, document in the band; document for all in Brazil
  • Signal under 18: closed

Fits adult companions

Australia's Head Terms list facial age estimation and photo ID matching among measures considered appropriate, plus an age signal from a party whose own assurance was appropriate; a bare self-declaration is not.[18] Not law yet: the EU KIDS Act proposal would let under-13s reach AI companions only through guardian tools.[21]

Streaming and media

A licensed catalogue often needs no check at all. Brazil exempts editorially controlled, licensed, music and literary services that offer child profiles plus parental supervision and blocking.[2] The draft puts mixed video platforms and explicit music or podcasts in the moderate band.[4]

  • Offer a child profile with rated content only.[2]
  • Add parental supervision and blocking.[2]
  • Review autoplay in California: AB 1709 names it and is not limited to social networks.[6]
  • Treat user uploads, live chat and comments as the social row.

Marketplaces and delivery

The trigger is the item, not the shop. In Brazil, whoever sells or intermediates alcohol, tobacco and vapes, weapons, fireworks, dependence-causing products, bets or loot boxes must verify age at sign-up, with purchases blocked by default and no release by self-declaration, or at purchase.[2] The official Q&A names delivery apps and marketplaces.[3] California's AB 1709 leaves out apps where users only trade or review.[6]

1Tag restricted items and block them by default

The rest of the catalogue stays open.

2Check at the first restricted checkout

Document, liveness and face match.

The document says 18 or over and no signal says under 18

Yes

Complete the order

Keep the yes, not the document or its data.

No

Remove the item

Offer a way to contest the age, as the decree requires.

Brazil flow.

Art. 24 of the decree bans keeping the document image or copy: deletion must be immediate and irreversible once the needed data is captured.[2] See our marketplaces page.

Betting and lotteries

No signal is enough here. Brazil lists gambling, betting and lotteries as prohibited for minors, asks for effective verification, bans and removes minors' accounts, and tells stores to block betting apps that are unauthorised or offer no age verification.[2] The law wants that verification at each access and bars self-declaration.[1] The draft accepts estimation only when complemented by verification.[4] In the EU, the Commission's guidelines, which are not law, recommend it for gambling.[20]

  • Ask for a document with liveness and face match before the account exists.
  • Refuse the account when signal or document says under 18.
  • Sweep existing accounts and remove those run by minors.[2]
  • Delete the document image at once in Brazil and keep the decision.[2]

Gambling licences add identity duties this page does not cover: see age verification for iGaming and our iGaming page.

Education apps

Most education apps should not add a face or document check. California's AB 2246 gives a business likely to be accessed by children a choice: estimate age with a level of certainty that fits the risk, or apply children's privacy protections to all consumers. The second needs no check, and the Act says age assurance must be proportionate.[7]

Default

Protect everyone as a child

  • High privacy settings for all
  • School roster or class code for access
  • Signal only where a law asks for it

Civil Code 1798.99.29

Exception

Adult or open areas

  • Public forums, tutor marketplaces
  • Check the adults, not the pupils
  • Open feeds follow the social row

Your risk assessment

In the US, the FTC's COPPA statement on data collected only to determine age covers general and mixed-audience operators, under conditions, not operators whose primary audience is children.[16] For adult learners and tutors, see our edtech page.

Apps that need nothing more

A calculator or a notes app should not ask for a selfie. AB 1856 "does not require the collection of additional personal information" beyond what section 1798.501 needs, and Colorado's 2028 law says nothing requires a developer to acquire age data "other than an age signal" and leaves out apps that process no personal data.[5][14] Brazil's draft gives low-risk services two examples: the signal and verifiable credentials.[4]

  • Request the signal where a law requires it. California bars requesting it where no law does.[5]
  • Keep one safe experience for all ages.
  • Use the signal only for age-appropriate experiences and legal duties, never for ads or analytics.[25]
  • Review again when you add chat, a feed, uploads, AI or purchases.

Pick a method per feature

Start with the lightest method that meets the rule. Brazil's draft recommends that sequence except where verification is mandatory.[4]

MethodWhat it tells youFitsDoes not fit
Protective defaultsNothing, by designEducation, utilities, any missing signalAdults-only products
Store signalAn age range, often declared: Google says it "may or may not have confirmed" TIER_A[24]Feeds in California, default profiles, low riskWeb, sideloaded apps, prohibited content in Brazil
Selfie estimate with a bandAn estimated age, not an identity[31]Dating, social and AI gates away from the thresholdBetting or prohibited goods in Brazil, when used alone
Document, liveness and face matchA date of birth bound to the personBetting, alcohol, loot boxes, borderline estimatesLow-risk apps

Missing or declared answers: see what to do when there is no age signal. Methods compared.

Show us your age gate

We map your features to the signal, the selfie and the document. Get a demo of age estimation in your app, or try demos.didit.me.

How Didit helps: two workflows and your own threshold

Didit is the age verification your app runs when the signal is missing or not enough. It does not read, replace or receive Google's or Apple's answer: your app starts a session with your own threshold.

FeatureDidit setupData setting
Dating, social and AI gatesAdaptive Age Verification: age estimation with liveness, document only inside the band you set[30]ISO/IEC 27566-1 template: a yes or no, no personal data, lowest minimum age 13, fixed at creation[30]
Brazil: betting, loot boxes, alcohol, sexual contentID Verification with liveness and face match for every user who asksWithhold fields, then delete the session after the result[36]. Counsel decides when a repeat access needs a new session
Several gates in one appYour own threshold as metadata, and a branch on it[33]Never the store signal in metadata

In a workflow, age estimation costs $0.10 and the ID Verification it triggers inside the band adds $0.15; the fallback face match is not billed (pricing).[32][30] With the fallback on, an estimate below the band declines, so start the band below your age limit.[34] Age limits set per country follow the document's issuing country, not the user's location.[30]

Data is stored until the retention window you set unless you delete it, and face search, on by default in the Age Estimation step, keeps no face template for that step once turned off.[36][35] Make both choices before launch in Brazil, with counsel. See it run.

Didit console, age estimation result with sample data: the estimated age from the selfie next to the age on the document

The age estimation result in the Didit console (sample data): the estimate from the selfie next to the age on the document.

Didit provides

  • Selfie age estimate with liveness
  • Document step-up, face matched to the selfie
  • Biometric consent step for Texas, Illinois and Colorado
  • Signed webhook with the decision

Stays with you

  • Google's and Apple's signals
  • The legal reading per market
  • Which feature to gate, at what age
  • Defaults, moderation and appeals

Gate the feature, not the whole app

See the selfie estimate and document step-up on your features.

Get a demoStart freeRead the docs

Key takeaways

  • A law, your terms or your risk triggers a check, not the category.
  • Utilities, most education apps and licensed catalogues can usually stop at the signal.
  • In Brazil, betting, reachable loot boxes, prohibited goods and sexual content need verification, not estimation alone.
  • In Alabama, Utah and Brazil your own check can only tighten.

Frequently asked questions

What does age assurance mean?

It is the umbrella term for ways to verify, estimate or infer a user's age or age range. Brazil's decree and Australia's codes use it broadly.[2][18]

What is the difference between age assurance and age verification?

Verification is one kind of age assurance. In Brazil it means a procedure with a high degree of reliability, while an estimate predicts the age from a face without identifying the person.[2][31]

Do all apps need age assurance?

No. California makes developers in its scope request the signal from 2027 but does not require extra personal information, and most low-risk apps can stop there.[5]

Do games need to verify age in Brazil?

Not every game. The official Q&A says so. Games with loot boxes must verify age unless they offer a version without them or keep the feature fully restricted by default.[3][2]

Do dating apps have to check age?

Apple rates apps with dating features 18+, and Brazil requires verification for apps mainly for sexual encounters. Elsewhere we found no dating-specific rule, so it is a risk decision to confirm with counsel.[27][2]

Do AI companion apps have to check age?

In Australia, AI chatbots that can generate age-restricted material without appropriate safeguards have to do age checks. In Brazil, sexually explicit AI dialogue is treated as pornography, which needs the provider's own verification.[17][2]

Is a store age signal enough for betting?

No. Brazil requires effective verification for betting and ANPD's draft guide asks for it regardless of the age signal received.[2][4]

Can I gate only one feature?

Yes. Brazil's loot box rule and its checkout rule are feature gates, and California's AB 1709 lets under-16s keep an account without the addictive feature.[2][6]

Is it safe to do online age verification?

It depends on what is collected and kept: ask which fields return, how long images stay and whether a face template is stored. In Didit you choose returned fields, retention and deletion.[36]

Can I verify age without ID?

You can estimate it: a selfie estimate needs no document. It is not proof of age, so use a document near the threshold and wherever a law asks for verification.[31][4]

Sources

  1. Lei 15.211 of 17 September 2025 (ECA Digital), Planalto, Arts. 1, 9, 14, 20, 21 and 24.
  2. Decreto 12.880 of 18 March 2026, Planalto, Arts. 2, 14 to 19 and 21 to 27.
  3. ECA Digital questions and answers, ANPD, Ministry of Justice and Secom, 30 July 2026.
  4. ECA Digital page, with the enforcement stages and the preliminary version (May 2026 draft) of the guide "Mecanismos de Aferição de Idade", ANPD. The guide is a draft, not final guidance.
  5. AB 1856, Chapter 184, Statutes of 2026 (amends AB 1043, the Digital Age Assurance Act), California Legislative Information.
  6. AB 1709, Chapter 183, Statutes of 2026, California Legislative Information, Bus. and Prof. Code 22682 to 22686.
  7. AB 2246, Chapter 187, Statutes of 2026 (Age-Appropriate Design Code), California Legislative Information, Civ. Code 1798.99.28 onwards.
  8. Health and Safety Code 27001 (SB 976), California Legislative Information.
  9. SB 976 rulemaking status, California Attorney General.
  10. HB 161, App Store Accountability Act, enrolled text, Alabama Legislature, s. 9.
  11. Utah Code 13-76-202, Utah State Legislature.
  12. SB 2420, App Store Accountability Act, Texas Legislature, Bus. and Com. Code 121.052.
  13. Act 185 of 2026 (HB 977), R.S. 51:1771 to 1775, Louisiana Legislature.
  14. SB 26-051, Chapter 343, Colorado General Assembly.
  15. Public Act 104-0664, Children's Online Social Media Safety Act, Illinois General Assembly.
  16. COPPA enforcement policy statement on age verification, 25 February 2026, Federal Trade Commission.
  17. Age-Restricted Material Codes fact sheet, April 2026, eSafety Commissioner, Australia.
  18. Register of industry codes: Head Terms and Schedule 7, eSafety Commissioner, Australia.
  19. Social Media Minimum Age regulatory guidance, September 2025, eSafety Commissioner, Australia.
  20. Guidelines on the protection of minors under the Digital Services Act, 14 July 2025, European Commission.
  21. EU KIDS Act, COM(2026) 681, proposal of 17 September 2026, EUR-Lex.
  22. How age checks at 16 can work, Ofcom, United Kingdom.
  23. Regional age-related protections on Play, page of 18 November 2026, Android Developers.
  24. Understand age signals responses, Android Developers.
  25. Play Age Signals API policy, Google Play Console Help.
  26. Changes to Google Play for upcoming app store bills for users in applicable US states, Google Play Console Help.
  27. New age rating content descriptors in App Store Connect, 2 November 2026, Apple Developer.
  28. Age requirements for apps distributed in Brazil, Australia, Singapore, Utah and Louisiana, 24 February 2026, Apple Developer.
  29. New requirements for social media apps in Australia, 8 December 2025, Apple Developer.
  30. Age estimation overview, Didit documentation.
  31. Age estimation API, Didit documentation.
  32. Pricing, Didit documentation.
  33. Workflows, Didit documentation.
  34. Age estimation warnings, Didit documentation.
  35. Feature configs, Didit documentation.
  36. Data retention, Didit documentation.
  37. Biometric consent, Didit documentation.

When your row asks for more than the signal, see Didit age verification.

Find your row, then see it run

Get a demo of age estimation in your app, with your threshold and band.

Get a demoStart free

身份与欺诈基础设施。

一个 API 即可实现 KYC、KYB、交易监控和钱包筛选。5 分钟即可集成。

让 AI 总结此页面