Changelog
Every change to the register, newest first, with how it was detected and its source. A gate status, a vendor or a retention quotation only ever changes by hand, after a human has read the source. Subscribe: RSS. Each service entry also has its own feed.
- — 4chan — seed: Record created from 4chan’s own Rules, FAQ and Legal pages (read in the browser on 18 October 2026) and Ofcom’s 4chan investigation page, its news pages of 13 October 2025 and 19 March 2026 and its June 2026 industry bulletin, with research/01 row 57 as the starting point only. The seed had the £520,000 fine as “from search summary, not fetched”; the figures are now from Ofcom’s pages. seed · source
- — Amazon Prime Video — seed: Record created from Amazon’s UK Prime Video help pages (Parental Controls on Prime Video; Maturity Ratings; Set up Restrictions on Prime Video on Web; Set up a Prime Video PIN on iOS) and the Prime Video Terms of Use, read 18 October 2026, with research/01 row 32 as the starting point only. No age check is described on any of them. seed · source
- — Apple (device-level age check and Declared Age Range API) — retention.source_date, retention.last_verified: Apple’s adult-confirmation page re-read 18 October 2026: Published Date now 18 September 2026; “kept for a short period of time” unchanged. Only the retention source was re-read, so the vendor record’s own last_verified stays at 6 September 2026. manual · source
- — Apple devices (UK account age confirmation) — open_questions, sources: Noted Ofcom’s 22 September 2026 investigation into Aylo’s use of Apple’s age-check signals. Apple’s own pages were not re-read; last_verified unchanged. ofcom · source
- — Apple devices (UK account age confirmation) — sources, methods.source_date: Both Apple pages re-read 18 October 2026. The adult-confirmation page now shows Published Date 18 September 2026; the card (“your credit card isn’t stored”) and ID (“kept for a short period of time”) wording is unchanged. New on the page: passport scanning listed for Australia, Singapore, the UK and Texas, and Apple Wallet Digital ID from a US passport accepted. The age-requirements page is unchanged (29 April 2026). (was: Apple adult-confirmation page dated 5 June 2026) manual · source
- — Bluesky — methods: Bluesky’s post re-read 18 October 2026 (unchanged, 10 July 2025): it names only credit card verification and face scans through KWS. The ID and face-scan rows now cite KWS’s own developer page, which names Veriff for ID scans (with a selfie in the UK), Yoti for face scans and Stripe for cards, and lists debit cards as unavailable in the UK. No retention wording on either page. (was: ID row sourced to Bluesky’s post; face-scan row secondary (press)) manual · source
- — Bumble — seed: Record created from Bumble’s help centre, Terms and Conditions of Use (17 June 2026) and privacy policy read 18 October 2026, with research/01 row 38 as the starting point only. support.bumble.com returns 403 to this environment’s fetches and browser; the articles were read through the help centre’s own article API, which serves the same articles. The seed’s Yoti attribution (press) is on no Bumble page and is recorded as an open question. seed · source
- — Character.AI — seed: Record created from Character.AI’s own help centre (What is Age Assurance?; Age Assurance FAQ; Age Assurance: What you need to know; What happens to my data during age assurance?; How do I verify my age?; Age Assurance Troubleshooting; Why am I not seeing the Age Assurance pop-up?; Why is Character.AI asking about my birthday?; Important Changes for Teens) and its blog posts of 29 October and 21 November 2025, read 18 October 2026, with research/01 row 48 as the starting point only. The seed’s source was press; every fact here is from Character.AI’s pages. seed · source
- — ChatGPT (OpenAI) — methods.source_date, gate.notes, summary, user_facing, open_questions: Help page re-read 18 October 2026 (updated about 14 October). Persona 7-day deletion wording unchanged. The page no longer states a 60-day Italian deadline and now says Persona may ask Italian users for a selfie, an ID or both; decline wording is now “Age prediction does not require you to verify your age to keep using ChatGPT with teen protections.” UK status still not stated, so gate status stays unknown. (was: Page dated about 30 August 2026; gate note said Italy had a 60-day deadline; decline wording “You do not need to verify your age to keep using ChatGPT with teen protections.”) manual · source
- — Crunchyroll — seed: Record created from Crunchyroll’s own help centre (How to enable mature content?; Understanding Content Ratings and Restrictions; How to set up Content Restrictions; How do I set up and manage a teen profile?; Why am I being asked to verify my age?) and its Terms of Service, read in the browser on 18 October 2026, with research/01 row 35 as the starting point only. An earlier batch on the same day skipped Crunchyroll because the help centre could not be read. seed · source
- — Depop — seed: Record created from Depop’s own help pages (How old do I have to be to use Depop?; Why does Depop need my date of birth?; Tips for parents and guardians), its Terms of Service and its privacy policy, read 18 October 2026, with research/01 row 55 (“N (13+ browse, 18+ sell)”, secondary source) as the starting point only. seed · source
- — Discord — methods, sources: Discord rewrote its UK article on 22 September 2026 (now “How Teens Experience Discord in the UK”) and moved the methods and retention wording to “How to Confirm Your Age Group on Discord” (24 September 2026). Rows re-quoted from that page: video selfie, ID scan, credit card (new), Google Wallet passport (no longer a test; UK passports listed), app-store age range (new) and AgeKey (new). (was: facial_age_estimation: “Video selfies for facial age estimation are processed entirely on-device and never stored.” (UK help page dated 28 July 2026) | photo_id: “Identity documents that go directly to k-ID are never seen by Discord. They are deleted immediately after age confirmation.” (UK help page dated 28 July 2026) | digital_id_wallet: “You can confirm your age group by using a Passport ID already verified and stored in your Google Wallet. We don’t get your Passport information, just your age.” (UK help page dated 28 July 2026)) manual · source
- — Discord — open_questions: Resolved from Discord’s own vendors page (September 2026): k-ID orchestrates and “The image of your ID is sent through a secure portal to Veratad, one of k-ID’s trusted third-party providers.” (was: Vendor conflict on the ID route: Discord's own help page (checked 7 September 2026) says identity documents go directly to k-ID; February 2026 press reported Veratad. Discord's page never names Veratad.) manual · source
- — Discord — gate.notes, summary, user_facing: Discord blog of 22 September 2026: account-signal age groups launched worldwide; “The age estimation model is not currently available for Discord users in the UK and Australia.” UK users keep the confirm-on-request approach. (was: Global rollout planned for the second half of 2026) manual · source
- — Disney+ — seed: Record created from the Disney+ Subscriber Agreement for the United Kingdom (updated 18 June 2026) and Disney+’s UK explainer “Parental Controls that fit your family, on Disney+” (revised 13 May 2026), read 18 October 2026, with research/01 row 34 as the starting point only. No age check is described on either; the help centre’s UK pages redirected to US versions. seed · source
- — EA (Electronic Arts) — seed: Record created from EA’s own help pages (age check; EA app age assurance; EA SPORTS FC Brazil page; parental controls; create a child account; upgrade to a teen account; create an EA Account), its UK User Agreement and Privacy and Cookie Policy, read 18 October 2026, with research/01 row 26 (“Unknown”) as the starting point only. seed · source
- — eBay — methods[other].source_url, sources, overview: eBay’s help page re-read 18 October 2026 (card wording unchanged). Verifymy’s two eBay pages had moved to /age-assurance/ (old addresses return 404); links updated and notes re-quoted. Still no retention statement for the eBay deployment. (was: Verifymy eBay pages under /age-verification-and-estimation/ (now 404)) manual · source
- — Epic Games and Fortnite — seed: Record created from Epic’s own pages (Cabined Accounts FAQ; Why is Epic Games asking for my date of birth again?; Parental Verification FAQ; How to correct the date of birth on your Epic account; Why do I need to verify my age on the Epic Games Store in Australia?; Parental Controls; Fortnite news of 7 December 2022) and Kids Web Services’ verification-methods and privacy-policy pages, read 18 October 2026, with research/01 row 25 as the starting point only. The seed had Fortnite’s gate as unverified; Epic’s pages describe an under-13 parental-consent flow and no UK 18+ check. seed · source
- — Epic Kids Web Services — retention, sources, open_questions: KWS privacy policy re-read 18 October 2026 (last updated 10 September 2026): quote made verbatim; the policy also says KWS records the method used “but we never store the verification details (e.g. your card details) themselves”. KWS developer page re-read: credit card (Stripe), ID scan (Veriff, with a selfie in the UK) and face scan (Yoti) available in the UK; debit cards not. (was: stores verified status with your hashed email address; retains for as long as necessary to fulfil the purposes) manual · source
- — Epic Kids Web Services — known_customers, sources: Epic Games and Fortnite added as a deployment from Epic’s own pages (read 18 October 2026): the Parental Verification FAQ says a parent may have to verify they are an adult by “placing an authorization or small refundable charge on a credit card, or checking some of your information against a public records database” and that adults “already … verified … through our subsidiary Kids Web Services Ltd” are not asked again. KWS’s developer page lists credit card (£0.05 authorisation), ID scan with a selfie and face scan as available in the UK. manual · source
- — Epic Kids Web Services — known_customers, sources: EA and Riot Games added as deployments from their own pages (read 18 October 2026). EA’s child-account page: a parent confirming they are an adult is “taken to our trusted partner Kids Web Services (KWS)”. Riot’s Age Verification page: a credit card or government ID “will be processed through Kids Web Services (KWS), a third-party service. Don't worry—Riot doesn't store or access any of this personal information, it's only used by KWS to confirm you're an adult.” Neither page states what KWS keeps for that deployment; the KWS privacy policy of 10 September 2026 is quoted on the service rows. manual · source
- — Facebook — seed: Record created from Facebook’s UK Help Centre pages and Meta newsroom posts read 18 October 2026, with research/01 row 8 (“as Instagram”) as the starting point only. Facebook’s own pages supply the Yoti partnership, the two routes and the 30-day deletion lines for the selfie image and the ID. seed · source
- — Feeld — seed: Record created from Feeld’s help articles, FAQ, UK Terms of Use (18 December 2025) and privacy policy read 18 October 2026, with research/01 row 40 as the starting point only. The seed’s retention line (press) is replaced by Feeld’s own wording. support.feeld.co returns 403 to ordinary fetches; the two articles were read through the help centre’s own article API, which serves the same articles. seed · source
- — Google (age inference and Google Account verification) — retention, methods, open_questions: Google Account help re-read 18 October 2026: retention quote made verbatim (selfie sentence; ID and card wording in the period field); digital ID in a wallet (Android only, not all regions) added to methods; Private ID named as the selfie partner. (was: ID deleted after your date of birth is successfully verified; selfie then deleted; card data retained as necessary to meet legal and regulatory requirements) manual · source
- — Google (age inference and Google Account verification) — known_customers, sources: Google Account and Google Search added as a deployment: Google’s own pages (How Google helps provide an age-appropriate experience; Make Google Search safer with SafeSearch; Access age-restricted content & features) read 18 October 2026 describe the account-level age estimate, the SafeSearch Filter default for accounts estimated under 18, and the ID, credit-card, selfie, email and digital-ID routes. manual · source
- — Google Account and Google Search — seed: Record created from Google’s own help pages (How Google helps provide an age-appropriate experience; Make Google Search safer with SafeSearch; Access age-restricted content & features) and YouTube’s age-estimation page, read 18 October 2026, with research/01 row 13 as the starting point only. Google Account verification was previously recorded only inside the YouTube record. seed · source
- — Grindr — sources.date, methods[photo_id].retention: Page re-read 18 October 2026; it now shows Updated 11 October 2026 and the deletion wording is unchanged. The same sentence covers documents, so the ID row now carries it instead of the absence line. (was: Page dated 31 August 2026; ID row carried no retention quote) manual · source
- — Hinge — seed: Record created from Hinge’s help pages read 18 October 2026 (Age Checks at Hinge, Face Check Scan, What is Selfie Verification, Availability, Help Center Change Log) and Hinge’s UK privacy policy, with research/01 row 37 as the starting point only. The help pages return 403 to ordinary fetches from this environment and were read through the help centre’s own article API, which serves the same articles. Hinge was previously folded into the Tinder record; it now has its own. seed · source
- — Instagram — seed: Record created from Meta’s Instagram help pages and newsroom posts read 18 October 2026, with research/01 rows 7 and 8 as the starting point only. The research row’s retention line (“Yoti shares an age estimation with Instagram”, press) is replaced by Instagram’s own 30-day deletion wording for the selfie image and the ID. seed · source
- — k-ID — retention.quote, methods, certifications, known_customers: k-ID privacy policy re-read 18 October 2026 (still dated 10 July 2026): quote made verbatim. ACCS registry re-read: Kidentify now holds an ACCS 1:2025 Age Assurance certificate alongside the Age Appropriate Design one. Discord’s vendors page (September 2026) names k-ID for credit card, facial age estimation, ID scan plus selfie and AgeKeys, with Veratad checking the document. (was: k-ID does NOT store the information that you provide to prove your age. All we store is the result of the verification process.) manual · source
- — Kick (streaming) — last_verified: FAQ re-read 18 October 2026: still dated 21 May 2026, all quoted wording unchanged; the FAQ still does not say which methods UK viewers receive or when the UK gate began. manual · source
- — Minecraft — seed: Record created from Minecraft’s help article on Java Edition chat access, the Xbox Support age-assurance page it links to, and Microsoft’s account age-assurance FAQ and how-to page (both dated 24 August 2026), read 18 October 2026, with research/01 row 24 as the starting point only. The seed’s retention line (press) is replaced by Microsoft’s own wording. seed · source
- — Mumsnet — seed: Record created from Mumsnet’s own Terms of Use, privacy policy and a staff reply of 31 July 2025 in its Site Stuff forum, read 18 October 2026, with research/01 row 56 (“N found”) as the starting point only. No age check is described on any of them. seed · source
- — Netflix — seed: Record created from Netflix’s UK help pages (How to verify your age; Parental controls on Netflix; Maturity ratings for TV shows and movies; How to set profile maturity ratings or block titles; How to add, edit, or remove a profile PIN; How to manage your account-level PIN) and its Terms of Use, read 18 October 2026, with research/01 row 33 as the starting point only. No age check is described on any of them. seed · source
- — Nintendo — seed: Record created from Nintendo’s UK support pages (Nintendo Account overview, under-16 accounts, child accounts, Parental Controls overview, GameChat guide and FAQ, GameChat guide for parents) read 18 October 2026, with research/01 row 21 as the starting point only. No age check is described on any of them. seed · source
- — Ondato — retention.quote, known_customers: OnAge page re-read 18 October 2026 (page modified 9 September 2026): quote made verbatim including Ondato’s own quotation marks. OnlyFans case study moved to /customer-success-stories/ (published 18 May 2026); link updated. ACCS registry entry re-read: two IEEE 2089.1 certificates, valid to 11 Feb and 8 Apr 2028. (was: deletes biometric data instantly. Only the result — old enough or not old enough — remains) manual · source
- — OneID — last_verified, open_questions: OneID age-verification, about and news pages re-read 18 October 2026: “No long-term personal data retention” unchanged; DVSTF and FCA statements unchanged; still no named UK consumer-service customer. manual · source
- — OnlyFans — methods, sources, open_questions: Facial-estimation row re-sourced to Ofcom’s 27 March 2025 decision (live selfie, challenge age) and Yoti’s own OnlyFans case study for the provider; Ondato case study link updated to its new address (18 May 2026). Still no retention statement from OnlyFans. (was: Facial-estimation row secondary (oursaferschools.co.uk); Ondato case study at /case-studies/) manual
- — Persona — retention.quote, certifications, known_customers, open_questions: Persona’s Reddit FAQ re-read 18 October 2026: the “within 3 days” wording is gone; the page now says deletion is immediate after the verification process ends. ACCS registry re-read: Persona Identities Inc holds an age-estimation certificate valid to 12 November 2026. OpenAI’s help page (14 October 2026) names Persona, replacing the press source for that customer. (was: Persona deletes all of your personal information including any images within 3 days.) manual · source
- — Persona — known_customers, sources: Character.AI added as a deployment from its own help pages (read 18 October 2026): “Character.AI has partnered with Persona, a trusted identity assurance provider, to handle all age assurance attempts.” Selfie first, government ID only if the selfie is inconclusive; Character.AI says “Any biometric data or ID submitted is automatically and permanently deleted within 7 days.” That figure is Character.AI’s statement, not a Persona page. manual · source
- — Pinterest — seed: Record created from Pinterest’s own help pages (Age requirements and verification; Teen safety options) and privacy policy, read 18 October 2026, with research/01 row 14 as the starting point only. The seed had Pinterest as “Unknown”; Pinterest’s own page describes a selfie or ID-plus-selfie check through a third party for age changes and age-related deactivations. seed · source
- — PlayStation — last_verified: FAQ re-read 18 October 2026: still “Updated June 1, 2026”; Yoti facial-geometry deletion wording unchanged; applies to adult accounts registered in the UK or Ireland. manual · source
- — Pornhub, YouPorn, RedTube (Aylo) — enforcement, summary, open_questions, methods[other].notes: Ofcom investigation opened 22 September 2026 (published 23 September) into the Apple-signal age assurance process, sections 12 and 36. Service pages were not re-read; last_verified unchanged. (was: No regulator action recorded) ofcom
- — Pornhub, YouPorn, RedTube (Aylo) — methods[other], sources: Aylo’s newsroom page re-read 18 October 2026. The 24 July 2025 method list (email-based age estimation, credit card, mobile network operator, open banking and, from 5 August 2025, Google Wallet) is unchanged. The page is a single running statement updated 23 September 2026 and carries Aylo’s own 5 May 2026 release on the Apple route, so that row is now primary. The page still publishes no retention wording for any method. (was: Apple device-check row sourced to press (Biometric Update, 8 May 2026), secondary) manual
- — Reddit — methods.retention: Persona deletes all of your personal information, including any images if collected, immediately after the verification process ends. (Persona’s Reddit FAQ re-read 18 October 2026; the page carries no date.) (was: Persona deletes all of your personal information including any images within 3 days.) manual · source
- — Reddit — methods, sources, user_facing.official_help_url, open_questions: Reddit’s current help pages read 18 October 2026: UK article (24 June 2026) and “Why is Reddit asking for my age?” (6 October 2026). Added the app-store age-range route and Reddit’s own age-prediction signals as rows; Reddit’s wording that it stores only verified age and status recorded in notes. (was: Official help link was a Reddit article that now returns “page doesn’t exist”) manual · source
- — Replika — seed: Record created from Replika’s own Privacy Policy, Terms of Service, help article and 2023 blog post, read 18 October 2026, and the European Data Protection Board’s report of the Italian authority’s decision of 10 April 2025, with research/01 row 49 (“Unknown”; Italian fine only) as the starting point only. seed · source
- — Riot Games — seed: Record created from Riot Games’ own support pages (Age Verification; Riot Games Parent Portal; Parents of Underage Players FAQ and Contact), its Terms of Service and Privacy Notice, read 18 October 2026, with research/01 row 27 (“N (UK); Y in Brazil”, UK/Yoti claim unverified) as the starting point only. Riot’s page names Kids Web Services as the processor of credit-card and government-ID checks. seed · source
- — Roblox — methods[photo_id].retention, open_questions, enforcement.source_url: Roblox’s own biometric privacy notice read 18 October 2026: the 30-day period is Persona’s deletion of biometric data from the ID-plus-selfie identity check, separate from the immediate deletion of facial age estimation images and video stated in the newsroom post (re-read, unchanged). Enforcement letter now cited to Ofcom’s own page. (was: Verify the separate reported biometric-data retention period against the current help policy. It must not be treated as a contradiction of the image/video deletion statement without checking data type, scope and exceptions.) manual · source
- — Snapchat — last_verified, enforcement.source_url: Snap’s UK Online Safety Act FAQ re-read 18 October 2026: still no age-verification wording. The 12 March 2026 letter is now cited to Ofcom’s own page (the earlier secondary link returns 404). manual · source
- — Spotify — summary, user_facing, methods[photo_id].retention: Page re-read 18 October 2026 (undated, unchanged): its example is an 18+ music video; it does not say mature podcasts are gated, so that claim was removed. The Yoti deletion sentence covers both methods, so the ID row now carries it. (was: 18+ music videos and mature podcasts need a Yoti check) manual · source
- — Steam — methods, sources, open_questions: Valve’s own support page read 18 October 2026; row upgraded to primary with Valve’s wording on credit cards, debit cards and PCI-DSS. No retention period published. (was: Credit-card row secondary (Game Developer, 29 August 2025)) manual · source
- — Telegram — sources, methods.notes, open_questions: Added Telegram’s developer documentation as a primary source for the bot mechanism. The press source (Durov’s Code) was unreachable on 18 October 2026, so last_verified was not advanced. manual · source
- — Threads — seed: Record created from Meta’s Threads help pages (Confirming your age on Threads, video selfie, ID, Teen Accounts, About Threads, report a user under the minimum age) read 18 October 2026, with research/01 row 9 as the starting point only. The seed had Threads as “no UK-specific gate found”; Threads’ own pages describe a Yoti video selfie or ID check for birthday changes and age restrictions. seed · source
- — TikTok — open_questions, enforcement: Ofcom’s investigation page re-read 18 October 2026: it states the case opened 16 July 2026 and remains Open, with an update promised in October 2026 (not published as of 18 October). The 12 March 2026 letter is now cited to Ofcom’s own page. (was: Investigation opening date confirmed only via secondary reporting) manual · source
- — Tinder — methods, summary, retention, user_facing, open_questions: Tinder’s own help page “Age Checks at Tinder” (updated 13 August 2026) read 18 October 2026. Facial row upgraded to primary with Tinder’s retention wording (results 1 year; photos and audit images for the life of the account); ID Verification Check added with its 30-day redacted-copy wording. The three-month and hashed-photo figures were Hinge’s terms as reported by press and have been removed from the Tinder facts. (was: Whole service gated for UK users: facial estimation with FaceTec liveness using Match's in-house model. Match states photos are kept up to three months, then a hashed photo and result for one year; Hinge audit images may train models.) manual · source
- — Tinder — name, gate.notes, open_questions: Hinge now has its own record built from Hinge’s help pages (data/services/hinge.json); this record covers Tinder only. (was: Record named “Tinder and Hinge (Match Group)”, with Hinge’s February 2026 rollout noted from press and an open question that Hinge’s own page had not been read.) manual · source
- — Tumblr — methods.retention, methods[open_banking], summary, gate: Tumblr’s post re-read 18 October 2026 (unchanged, 6 May 2026). Quotes made verbatim to the single sentence the post carries. The post lists no methods, so the bank (ConnectID) row is now secondary, cited to press; first_observed widened to May 2026 because the post says the requirement was coming “soon”. (was: Face row quote “Facial scan data never leaves your device.”; ID row quote “Tumblr won't have access to the information you submit, just the final age-verification result; Documents are deleted after confirmation”; ConnectID row primary from Tumblr’s post; first_observed 6 May 2026) manual · source
- — Twitch — methods.retention.quote, open_questions: k-ID’s Twitch page re-read 18 October 2026; quote made verbatim (the page’s wording includes the vendor parenthesis and “used for facial age estimation”). The page offers no in-app ID route, only an appeal through Twitch Support with age documentation, so the open question is answered and recorded in the row notes. (was: Quote: “Twitch and k-ID do not store your face scan video selfies. The video selfie is analyzed entirely on your device and will never leave it.”; open question: Whether an ID alternative exists for users the scan cannot estimate) manual · source
- — Veratad — retention, known_customers, sources, open_questions: Veratad’s privacy policy (last update 3 March 2026) read 18 October 2026 and quoted; the product page still carries no retention statement. Discord’s own vendors page (September 2026) confirms Veratad checks ID documents for Discord through k-ID, answering the press-only question. ACCS entry re-read (VX platform v2.4, valid to 14 Apr 2029). (was: No retention statement found on the age-verification product page (checked 6 Sep 2026).) manual · source
- — VerifyMy — retention.quote, known_customers, sources, open_questions: Google policy re-read 18 October 2026 (text still “last updated June 2024”): quote made a full sentence. Verifymy’s eBay pages moved to /age-assurance/; links updated. ACCS entry re-read: IEEE 2089.1 (to 21 Sep 2027) and ACCS 1:2025 (to 20 May 2029) under the name VERIFYMY LIMITED. (was: eBay pages under /age-verification-and-estimation/ (now 404); quote fragment) manual · source
- — VerifyMy — known_customers, sources: Google Account and Google Search added as a deployment: Google’s verification page (read 18 October 2026) says “Google provides your email address to VerifyMy, a third-party partner that performs the verification” and “VerifyMy doesn’t save your email address or use it for any purpose except to verify your age.” The same route already appears under YouTube. manual · source
- — Vinted — seed: Record created from Vinted’s own help pages (Verify your age; Signing up on Vinted), its UK User Terms and Conditions and its privacy policy, read 18 October 2026, with research/01 row 54 (“N found”) as the starting point only. Vinted’s help centre describes a document check for accounts it asks to verify. seed · source
- — WhatsApp — seed: Record created from WhatsApp’s UK Terms of Service (effective 17 August 2026) and its help pages on the minimum age and parent-managed accounts, read 18 October 2026, with research/01 row 10 as the starting point only. No age check is described on any of them. seed · source
- — Wikipedia — last_verified: Wikimedia UK post re-read 18 October 2026 (published 9 January 2026, unchanged): the High Court challenge was dismissed and a Category 1 decision is expected in 2026; no age gating. manual · source
- — X (Twitter) and Grok — last_verified: X’s age assurance page re-read 18 October 2026: wording unchanged; it still names no vendor, publishes no retention period for any method and carries no page date. manual · source
- — Xbox — methods.retention, methods.notes, availability, sources.date: Xbox support page, Microsoft FAQ and how-to page re-read 18 October 2026 (Microsoft pages dated 24 August 2026). ID and facial rows now carry Microsoft’s current sentence “It is securely processed by our trusted age assurance partner and deleted after the process is complete.”; mobile-provider and credit-card rows are flagged as absent from the current FAQ. Xbox Wire wording unchanged. (was: ID row quoted the July 2025 Xbox Wire sentence “is protected with industry-standard encryption and is not stored or used for any other purpose.”; other rows carried the absence line) manual · source
- — Yoti — retention, certifications, accuracy_claims, regulatory_history, sources, open_questions: Yoti’s Age Verification Service privacy notice (23 February 2026) read 18 October 2026 and now the retention source. White paper link moved to Yoti’s CDN (same July 2025 document; wording “The images are not stored, shared, re-used or sold on…” confirmed). ACCS registry now lists Yoti Ltd with age-estimation and age-assurance certificates. AEPD fine now cited to the AEPD resolution itself and to Yoti’s 10 March 2026 response, replacing Wikipedia; the breakdown matches the resolution. (was: Retention quoted from the white paper PDF (link now 404): “Images are not stored, shared, re-used or sold on. Images are immediately and permanently deleted… we do not use them for our own learning or training purposes.” AEPD entry dated 25 March 2026 and cited to Wikipedia: “€950,000 fine: €500k unlawful biometric processing, €200k invalid consent, €250k excessive retention. Yoti rejects the decision; appeal pending.”) manual · source
- — Yoti — known_customers, sources: Instagram and Facebook now have index entries. Both are cited to Meta’s own pages: Meta’s 23 June 2022 post (“We are also partnering with Yoti”) and Facebook’s Help Centre (“Facebook contracts with Yoti, a third-party age verification service”). Yoti’s July 2025 fact sheet, which names Instagram and Facebook among its users, added as a source. (was: Instagram listed as a reported customer from a November 2024 press article, with no index entry to link to.) manual · source
- — Yoti — known_customers, sources: Feeld (its UK Terms of Use, 18 December 2025, name Yoti for facial estimation and mobile network operator checks), Threads (Meta’s Threads help page: “We partner with Yoti”) and Minecraft (Minecraft’s help page sends UK players to Microsoft’s account age verification, whose FAQ names Yoti and Verifymy as partners) added as customers with their own index entries. (was: Five named customers: Spotify, Xbox, PlayStation, Instagram, Facebook.) manual · source
- — YouTube — methods.retention, methods, enforcement.source_url: Google Account help, YouTube age-estimation page and Verifymy’s Google policy re-read 18 October 2026. Quotes made verbatim to full sentences; digital ID (Android only) added as a route; the UK is still on YouTube’s list of countries where age estimation is applied. The 12 March 2026 letter is now cited to Ofcom’s own page. (was: Card quote “retained as necessary to meet legal and regulatory requirements”; selfie quote “then deleted”) manual · source
- — YouTube — name: Google Account verification and Google Search now have their own record built from Google’s Account and Search help pages (data/services/google-account.json); this record covers YouTube only. Its method rows are unchanged because YouTube users verify through the same Google Account routes. (was: Record named “YouTube and Google Account”) manual · source
- — eBay — record: Added the eBay retail purchase record with separate credit-card and VerifyMyAge routes. manual · source
- — VerifyMy — known_customers: Linked the eBay deployment from VerifyMy’s own integration page. Existing Google retention wording remains scoped to Google; other vendor facts were not rechecked. manual · source
- — Xbox — summary, scope and current support: Read current Xbox and Microsoft support. Separated regional content and feature restrictions from the July 2025 UK announcement; kept historical methods qualified and owned games accessible. (was: UK users must verify age to keep voice and text chat, invites, LFG and clubs.) manual · source
- — Apple devices (UK account age confirmation) — gate.status, methods, retention, sources: Apple guidance confirms feature restrictions, account-information route and distinct card/ID retention treatment. (was: Whole-service gate and press-based method/retention claims) manual · source
- — ChatGPT (OpenAI) — gate: Aligned UK status with the existing source caveat. Removed the non-UK-specific January date from the UK first-observed field. No source was newly verified. manual · source
- — Discord — summary: Corrected the introduction to match the existing primary-source method records; source verification dates are unchanged. manual · source
- — Discord — methods: Re-read the UK help page and confirmed the recorded methods, k-ID routes and Google Wallet test. Updated method verification dates only. manual · source
- — Discord — methods.availability: Made the existing Google Wallet test limitation explicit across summaries and agent responses; source verification dates are unchanged. manual · source
- — Kick (streaming) — methods.availability: Exposed the existing jurisdiction limitation as a method availability field; source verification dates are unchanged. manual · source
- — Pornhub, YouPorn, RedTube (Aylo) — retention: Removed an account-access statement from the retention field because it does not describe data retention. manual
- — Reddit — methods.retention: Re-read Persona’s Reddit FAQ. The current period is recorded separately from July 2025 press reporting; only method verification dates were updated. manual · source
- — Roblox — methods.retention: Re-read the newsroom source. Limited the retention period to images and video, removed the unsupported holder classification, and moved the separate reported biometric-data period to the review queue. manual · source
- — ChatGPT (OpenAI) — methods: help page read in a browser (previously 403): the seven-day Persona deletion figure is now quoted from OpenAI's own page rather than press; age-prediction inference added as a method; UK rollout still not stated by OpenAI manual · source
- — Discord — methods: help page read in a browser (previously 403): face-scan and ID rows upgraded to primary with Discord's own retention wording; Google Wallet passport method added; ID vendor recorded as k-ID with the Veratad press claim kept as an open question manual · source
- — Kick (streaming) — methods: help page read in a browser (previously 403): all rows upgraded to primary with Kick's own wording, dated 21 May 2026; credit-card and AgeKey (reusable digital ID) methods added manual · source
- — X (Twitter) and Grok — methods: help page read in a browser (previously 403): rows upgraded to primary and vendor names removed from them because X names no vendor; phone-number and account-signal inference added; the 30-day retention figure moved to notes as press-only, so the rows now render the no-published-retention line manual · source
- — Apple devices (UK account age confirmation) — seed: record created from research/01 and research/02 seed · source
- — Bluesky — seed: record created from research/01 and research/02 seed · source
- — ChatGPT (OpenAI) — seed: record created from research/01 and research/02 seed · source
- — Discord — seed: record created from research/01 and research/02 seed · source
- — Grindr — seed: record created from research/01 and research/02 seed · source
- — Kick (streaming) — seed: record created from research/01 and research/02 seed · source
- — OnlyFans — seed: record created from research/01 and research/02 seed
- — PlayStation — seed: record created from research/01 and research/02 seed · source
- — Pornhub, YouPorn, RedTube (Aylo) — seed: record created from research/01 and research/02 seed
- — Reddit — seed: record created from research/01 and research/02 seed · source
- — Roblox — seed: record created from research/01 and research/02 seed · source
- — Snapchat — seed: record created from research/01 and research/02 seed · source
- — Spotify — seed: record created from research/01 and research/02 seed · source
- — Steam — seed: record created from research/01 and research/02 seed · source
- — Telegram — seed: record created from research/01 and research/02 seed · source
- — TikTok — seed: record created from research/01 and research/02 seed · source
- — Tinder — seed: record created from research/01 and research/02 seed · source
- — Tumblr — seed: record created from research/01 and research/02 seed · source
- — Twitch — seed: record created from research/01 and research/02 seed · source
- — Wikipedia — seed: record created from research/01 and research/02 seed · source
- — X (Twitter) and Grok — seed: record created from research/01 and research/02 seed · source
- — Xbox — seed: record created from research/01 and research/02 seed · source
- — YouTube — seed: record created from research/01 and research/02 seed · source
111 entries across 48 services and 10 vendors. Detection labels: seed = created from the September 2026 research; manual = a human edit with the source read directly; diff = a source page changed and a human confirmed it; ofcom = from an Ofcom or ICO publication; user_tip = sent through the correction form and then verified.