Research and response update — November 12, 2026. An IEEE Security & Privacy 2026 study by researchers from Georgia Tech and UC Irvine identified privacy and security concerns in online age-verification systems, including browser/device metadata and information flows involving Yoti. However, a much more serious allegation that Yoti sent users' facial images to third parties was challenged by Yoti, and on June 19 Yoti said the universities had removed that assertion from their news coverage. Readers should distinguish these separate claims. This story is about a contested research finding, not proof that Yoti suffered a data breach or sold face images.

What did the researchers actually study?

The research paper was “Papers, Please: A First Look at Age Verification on the Web,” by Shreyas Minocha, Isaac Sheridan, Harry Oppenheimer, Paul Pearce and Michael A. Specter, presented at IEEE Symposium on Security and Privacy 2026.

The researchers examined how online age checks were deployed in Texas, Georgia and New York, using web crawls conducted in August 2025. Texas and Georgia had relevant state mandates while New York was used for comparison.

Their research-lab summary says that they detected age-check services on over 1,500 sites in each of Georgia and Texas, and on over 500 sites in New York. Yoti appeared on more than 60% of the sites in which the team detected a service. These are observations within the study's methodology, not a verified global market-share estimate for every age-verification provider.

The team also reverse-engineered browser-side elements of Yoti's service to assess privacy and security behavior. It had no privileged access to Yoti's internal backend, which limits what its observations alone can prove.

What information did the paper find in verification flows?

The study documented that some age-verification implementations collected device and browser metadata, including technical attributes that could potentially help identify a device across sessions. The researchers' own summary describes the resulting fingerprinting risk.

Device fingerprinting means deriving a relatively recognizable profile from signals such as software configuration, browser capabilities and hardware characteristics. Those signals need not include a person's full name or photograph to pose a privacy concern.

The research also described fourth parties involved in processing particular checks, including providers used for payments, IP geolocation and document validation. A fourth party is a service contacted by a party already integrated into the original website; its existence does not by itself establish unlawful trading of private information.

The important issue is which precise data was sent to which service, under what purpose and retention rules. These differ across different methods of age assurance.

What was the issue involving Stripe?

According to the researchers' March summary and footnote, Yoti's credit-card-based age-check flow disclosed the first-party website being accessed to payment processor Stripe.

The authors said Yoti informed them that the issue had been fixed, but they had not independently verified that fix at the time their summary was published. That is a carefully limited claim about a particular payment flow and the visibility of the destination website.

It does not mean Stripe was proven to have received every user's facial image, nor that an identity-theft incident occurred. The previous GameLogNews article blurred these categories, implying unqualified sharing of highly sensitive files and an ongoing system compromise.

A provider might legitimately integrate a payment service while still raising concerns over whether all the shared metadata was necessary. Those questions deserve clear, evidence-based reporting.

Did Yoti send facial photographs to third parties?

That allegation was publicly disputed and corrected. Yoti's chief executive Robin Tombs published an open letter on May 22, updated June 19, 2026.

The company said that facial-image data used for facial age estimation is processed in Yoti's own systems, is not shared with external parties, and is deleted immediately after the check. These are the vendor's direct representations about its product.

In its June 19 update, Yoti stated that the universities had removed the claim about facial images being sent to third parties from their May 19 news coverage and asked outlets to correct the false version. It did not say all other research findings had necessarily been withdrawn or overturned.

That is why a responsible update should acknowledge the retraction of the facial-image-sharing accusation without making the opposite sweeping claim that every aspect of the researchers' browser-metadata and payment-flow analysis was disproven.

The original Yoti privacy-policy page also describes a distinction between facial age estimation and identity-document verification; these are different processing methods with potentially different data requirements. See Yoti's published age-verification privacy policy.

Was this a hack or a proven privacy violation?

No hack was established by the sources reviewed for this correction. The IEEE paper was primarily an examination of deployment, browser-side data handling and the strength of security claims. It should not be retold as evidence that attackers breached Yoti or that Yoti had been convicted of illegally selling customer records.

More On Age Verification Tech Under
Age Verification Tech Under hubGaming News coverageMore from Julian at GameLog

The authors raised concerns about possible circumvention of some client-side integrity protections. Their technical findings were derived from particular implementations and versions, not universal proof that every current Yoti check can be bypassed.

The company contested important aspects of the public narrative and pointed to certifications and auditing. Yoti's response should be included because allegations involving a named security vendor can have reputational consequences and because the disputed statement was materially corrected.

At the same time, a vendor's statement about its architecture is not automatically an independent validation of every privacy or security claim. An accurate article can present both the research and the company's response without equating them.

Why does this matter to PlayStation and social-media users?

Yoti provides age-verification technology used in services relevant to gaming and online platforms. The university press coverage mentioned customers including Sony PlayStation and Meta.

That does not show that each of those clients used the same Yoti verification method, browser code, payment integration, location or retention policy. Different products may call different services.

Players concerned about an age check should find out who processes the image or document, whether a result can be disclosed as a simple yes/no age threshold, and what privacy notice applies to that specific flow. An age-estimation selfie is not necessarily processed the same way as a scanned passport or card-based verification.

The broader policy question is whether web services can limit unnecessary data collection while still meeting legal requirements to protect minors. The Georgetown/Knight technical review considers those trade-offs across multiple methods rather than attributing every problem to one vendor.

Did the paper establish that age checks are ineffective everywhere?

The researchers reported low detected deployment rates among sites self-identifying as adult-oriented in the jurisdictions examined, with specific measurements of 13.7% and 14.8% for Georgia and Texas.

Those are study results with an important qualification: automated crawling can miss age-verification systems that only appear after a login, are loaded conditionally or use a method the crawler does not recognize. The authors expressly identified their findings as a lower bound.

It would therefore be misleading to say exactly that percentage of every US site was legally compliant, or that all age-verification technology fails. The data supports questions about deployment and privacy trade-offs, but cannot establish a universal pass/fail verdict.

Quick answers

Did researchers identify browser/device fingerprinting risks? Yes, in their IEEE 2026 study.

Did they raise concerns about the Stripe payment flow? Yes. Yoti said it had fixed the first-party-site disclosure; the researchers said they had not verified the fix.

Did Yoti admit sending selfies to Stripe or other third parties? No. Yoti denied that facial images leave its own systems, and reported that the universities corrected their earlier allegation.

Was a Yoti data breach proven? Not by this study or the cited statements.

Does this mean PlayStation leaked players' ID photos? No such conclusion is established by the sources.

Bottom line: The IEEE study raised significant, concrete questions about metadata collection and age-verification design, while a highly sensitive facial-image-sharing accusation was challenged and subsequently corrected. Both are necessary to understand the actual 2026 record.

Primary sources: Georgia Tech research team summary; IEEE paper citation and reproducibility artifacts; Yoti's May 22 letter and June 19 update; Yoti official privacy policy; Georgetown technical analysis of age assurance.

Sources and Context
Primary source: Georgia Tech SPDR Lab
Source date: May 31, 2026