MADRID — Spain’s data protection authority has imposed a total fine of €950,000 on Yoti Ltd, the British digital identity and age verification company, after determining that the company committed three separate violations of the General Data Protection Regulation (GDPR) in connection with the operation of its Digital ID application.
The decision, issued under file reference EXP202317887, was signed by Lorenzo Cotino Hueso, president of the Agencia Española de Protección de Datos (AEPD). The ruling provides a detailed examination of the regulatory obligations that apply to age verification providers operating in Spain.
The three penalties consist of €500,000 for unlawful processing of biometric data under Article 9 of the GDPR; €200,000 for obtaining invalid consent for research and development processing in violation of Article 7; and €250,000 for excessive data retention in breach of the storage limitation principle set out in Article 5.1(e). In addition to the financial penalties, the authority ordered Yoti to implement corrective measures within six months after the resolution becomes final.
Yoti Ltd, registered in the United Kingdom with tax identification number 08998951, provides age verification services used by platform operators across multiple markets. According to the resolution, all of the company’s verification methods — including facial age estimation, document-based verification, credit card checks, mobile number matching and the Digital ID application — are available for use in Spain. The company’s most recent published revenue figure, cited in the resolution as of March 2025, is €15,029,907, which the authority used as a reference point in determining proportionate and dissuasive penalties.
How Yoti’s technology works
The Digital ID application is the service at the center of the enforcement action. According to documentation submitted during the investigation, the application allows users to create a verified identity account by uploading a government-issued identity document and capturing a selfie image.
The technology uses deep neural networks to process the facial image. The image is converted into pixels treated as numerical values and analyzed through a layered network of mathematical nodes. A typical run through the system produces an estimated age in approximately one to 1.5 seconds.
Yoti describes its services to business clients as comprising eight verification methods. According to the company’s data protection impact assessment (DPIA), these include facial age estimation, verification through the Digital ID application, document identification, credit card verification, mobile number verification, database checks, electronic identity systems used in Switzerland, Denmark and Finland, and a U.S. mobile driver’s license option. When these services are offered on a software-as-a-service basis, client companies act as data controllers while Yoti acts as a processor. Within the Digital ID application itself, however, Yoti acts as the controller.
The facial age estimation model was trained using 12 age range categories (0-1, 2-3, 4-6, 7-9, 10-12, 13-15, 16-17, 18-24, 25-29, 30-39, 40-49 and 50-60), four gender groupings, and three skin tone groups based on the Fitzpatrick scale, producing 144 demographic combinations. According to a company white paper referenced in the resolution and updated in September 2024, the model demonstrated accuracy within 1.28 years across gender and skin tone categories.
Training images were collected through an online portal that required adult consent, as well as through a South African family welfare organization, Be In Touch, working with schools. The United Kingdom’s Information Commissioner’s Office, which previously included Yoti in a regulatory sandbox program, advised against the South African collection method due to potential data protection implications.
The Digital ID application also applies age restrictions based on jurisdiction. According to Yoti, “the Digital ID app cannot be used by persons under the digital age of consent, i.e. 13 years in the United Kingdom and 14 years in Spain.” During account creation the application detects a user’s location and, in Spain, presents two options: “I am 14 or over”or “I am 13 or under.” The registration process continues only if the user selects the first option. No technical mechanism verifies the accuracy of the declaration.
For repeated verification, Yoti implemented a cookie-based age token system. These tokens remain valid for 30 days, allowing users who have verified their age once to reuse the result across participating platforms. The company also provides an “age account” feature that stores tokens in a username-and-password account accessible across devices.
First violation: biometric special category data
The AEPD’s primary finding concerns the processing of biometric data without a valid legal basis under Article 9 of the GDPR. The regulation prohibits the processing of special category data — including biometric data used for identification — unless specific exemptions apply.
Yoti maintained during the investigation that the facial scans generated by its system should not be considered special category biometric data because they are intended to authenticate users rather than uniquely identify them. The authority rejected this interpretation.
According to the resolution, data qualifies as biometric special category data under Article 4.14 of the GDPR when it relates to physical or behavioral characteristics of an individual, is used to confirm unique identification and undergoes specific technical processing to generate biometric templates. The AEPD determined that Yoti’s system meets all three criteria.
The authority found that the facial scan produces a biometric template stored while the user account remains active. When users modify their PIN or recover their account, the system captures a new facial scan and compares it with the stored template through a 1:1 matching process.
According to the decision, “despite repeatedly asserting — both during account creation and in the privacy policy — that the purpose of processing the biometric facial pattern is to guarantee user identification, Yoti does not consider itself to be processing special category personal data,” a position the authority described as demonstrating “particular negligence.”
The fine for this violation was set at €500,000. The authority cited the involvement of minors and the international processing of data — including servers outside the European Union — as aggravating factors.
For transfers between the United Kingdom and India, where Yoti operates a Security Centre providing manual verification support, the company relies on EU standard contractual clauses with a UK addendum. According to the DPIA, personnel at this center can access document images and selfies through remote connections to UK servers using “thin terminals,” while no other staff outside the center can view the information. The AEPD noted that the cross-border dimension further limits users’ practical control over their data.
Second violation: pre-ticked consent boxes for R&D
The second violation concerns the mechanism used to obtain user consent for internal research and development.
According to the investigation, the application displayed a pre-selected checkbox allowing users’ biometric data to be used to train and improve Yoti’s facial age estimation algorithms unless users manually deselected the option.
Yoti’s documentation confirms this design. The company stated, “In the Digital ID app, the default value is that data can be used for R&D. Yoti has taken steps to make this clear to users. Users can opt out, preventing their data being used for R&D, by using the app settings.”
The AEPD determined that this approach does not meet GDPR requirements. Article 4.11 defines consent as “any freely given, specific, informed and unambiguous indication of the data subject’s wishes” expressed through a clear affirmative action. Pre-ticked checkboxes do not constitute such an action.
The authority cited the European Data Protection Board’s Guidelines 05/2020, which state that consent obtained through default settings cannot be considered valid even if users later have the ability to withdraw it. The resolution notes: “Yoti consciously notes that consent granted by default can be revoked, without taking into account that there should not be a subsequent revocation at the data subject’s request, but rather that consent should be obtained in accordance with the safeguards and guarantees established by the GDPR.”
According to the DPIA, the data used for research processing may include facial images with timestamps, dates of birth derived from identity documents, gender information, document type details, country codes, video and audio recordings, device information, behavioral data, health-related information and race or ethnicity estimates derived from the Fitzpatrick scale. Data from users aged 13 to 18 is also included.
The fine for this violation was set at €200,000, with the authority again citing the involvement of minors and processing on servers outside the EU as aggravating circumstances.
Third violation: retention periods beyond stated purposes
The third infringement relates to the retention of personal data — including biometric data — for longer than necessary.
According to Yoti’s DPIA, Digital ID data and age tokens may be retained while a user account remains active or for three years after the last activity. The biometric facial template is stored throughout that period. The AEPD found this disproportionate.
The authority determined that the liveness check, which verifies that a real person is present during registration, is completed during account creation. Once this verification occurs, the purpose of the biometric capture is fulfilled. Retaining the biometric pattern beyond that moment cannot be justified by reference to that completed purpose.
The resolution also noted that additional uses of the biometric template — such as PIN modification or account recovery — may never occur during the account’s lifetime, meaning the storage of biometric data for possible future events fails to meet the storage limitation principle.
The authority also identified concerns regarding geolocation data. According to Yoti, the company collects users’ country code, city and state derived from their IP address and retains this information for five years. The stated purpose is to determine which jurisdiction’s age restrictions apply. The AEPD concluded that once jurisdiction is determined during account creation, extended retention of the location data is unnecessary.
Another retention issue involved fraudulent identity documents. Yoti indicated that documents identified as fraudulent may be stored for up to two years to train fraud detection systems. The authority determined that improving software constitutes a separate purpose not directly related to the original identity verification objective.
Video recordings created during liveness checks were also examined. The company’s terms state that such recordings “will be permanently deleted within 30 days of the date it was recorded, unless we are required to retain it for regulatory reasons.” The authority concluded that once liveness is confirmed, retention beyond that moment exceeds the legitimate purpose of the recording.
The fine for this infringement was set at €250,000, reflecting the large number of affected users and the involvement of special category data.
Corrective measures and timeline
The AEPD ordered Yoti to implement three corrective measures within six months after the decision becomes final:
• Demonstrate that the processing of biometric special category data complies with GDPR requirements.
• Demonstrate that consent-based processing meets the standards established by the regulation.
• Demonstrate that personal data retention is limited to the period strictly necessary for each processing purpose under Article 5.1(e).
The decision becomes final once the one-month period for filing an administrative appeal before the AEPD presidency has passed without action, or once the resolution is formally notified if no appeal is filed. Yoti may also challenge the ruling before the Contentious-Administrative Chamber of the National Court within two months of notification.
Failure to comply with the corrective measures could constitute a separate administrative violation under Articles 83.5 and 83.6 of the GDPR, potentially resulting in further enforcement proceedings.
Regulatory context
The Yoti ruling forms part of a broader series of enforcement actions by Spain’s data protection authority. The AEPD previously imposed a €500,000 fine on FC Barcelona for deficiencies in a data protection impact assessment related to biometric facial and voice data from approximately 143,000 members. The authority also issued a €1.8 million fine against airport operator AENA over the deployment of facial recognition systems, and a €1.8 million penalty against Informa D&B for processing personal data without a valid legal basis.
The decision also references the European Data Protection Board’s Statement 1/2025, adopted in February 2025, which outlines ten principles for GDPR-compliant age assurance systems. These include requirements that age verification technologies use the least intrusive methods available, avoid enabling tracking or profiling and implement short retention periods.
The ruling highlights ongoing differences in how European regulators interpret biometric data rules. While previous guidance from the United Kingdom’s Information Commissioner’s Office indicated that facial age estimation may fall outside biometric identification rules when used only for categorization, Spain’s AEPD concluded that persistent facial templates used for matching operations constitute biometric processing under Article 9.
The decision underscores increasing scrutiny of age verification technologies across Europe as regulators examine both their effectiveness and the privacy implications of the systems used to implement them.
Read More »