In this blog series, our CEO Robin Tombs will be sharing his experience, whilst focusing on major themes, news and issues in the world of identity verification and age assurance.
Digital IDs have been a hot topic this month. Robin spoke to Times Radio about digital IDs and shares some more thoughts on them below. He also addresses some inaccurate claims about facial age estimation.
Digital IDs are a focus for the new government
The new UK government is clear that individuals should be able to use certified digital IDs to prove who they are with more businesses. This would help people to complete age and identity checks in a secure, easy and digital way.
This is something the government stated in the King’s Speech, which announced the new Digital Information and Smart Data Bill. This Bill aims to enable new, “innovative” uses of data to help boost the economy.
Under this new legislation, digital verification services would be established, including digital identity products to help the public quickly and securely prove who they are, and share key information about themselves as they use online services in their everyday life. Leveraging digital verification will be increasingly important over the lifetime of the new UK government to improve productivity and reduce fraud.
On the topic of digital IDs, I was pleased to speak to Kait Borsay on Times Radio, talking about how digital IDs work and addressing some privacy concerns.
During the interview, I spoke about the importance of giving people a choice in how they prove who they are. They could use a certified digital ID or continue to use a physical ID and paper documents. No digital ID should be compulsory. This is something Susannah Copson, from Big Brother Watch, echoed during the interview.
We also discussed the importance of privacy and security. Any certified digital ID must have privacy throughout. At Yoti, we have championed this and focused on getting privacy and security right from day one. If you’d like to learn more about how we built our Digital ID with privacy at the core, this blog is worth a read.
Millions of Digital ID users in the UK
A recent Daily Mail article discusses the new government’s plan to introduce laws which will accelerate the use of reusable digital IDs. Great to see the article show how individuals can use their Yoti to share a data minimised Over 15 or Over 18 without having to show all their passport or driving licence details.
Nearly five million individuals in the UK have downloaded Yoti ID, Post Office ID or Lloyds Bank Smart ID. All of these apps can be accepted by businesses for government certified proof of identity (for right to work, right to rent and DBS checks), and for proof of age, approved by the UK’s PASS scheme.
There are some sceptics about digital IDs but much of the criticism ignores how the Digital Identity and Attributes Trust Framework (UKDIATF) has been designed, and how smart regulation can protect the UK government or digital ID providers trying to create a surveillance system.
The Daily Mail notes “Ex-Tory chairman Sir Jake Berry branded digital IDs ‘a creepy state plan to track you from the cradle to the grave’.”
But that’s not how our Digital IDs work. And nearly five million British people – some of whom have used Yoti for over 7 years – trust this. They know that after registration, we can’t see their personal information, how they are using the app or who they are sharing their details with.
Our Digital ID is designed with privacy throughout. Each individual user has control of their unique, single encryption key. This is secured in their phone and allows only them to unlock and share their details. Because we can’t see their details, we can never spam Yoti users with emails or texts!
Individuals in the UK are free to choose to use a free decentralised reusable Digital ID from Yoti, Post Office or Lloyds Bank for as long (or as short a time) as they wish – without fear of being tracked from cradle to grave.
Response to inaccurate claims about facial age estimation
In June 2024, four researchers – Stardust, Obeid, McKee and Angus – at Queenstown University of Technology published a paper titled: Mandatory age verification for pornography access: Why it can’t and won’t save the children.
They made some bold and negative claims about facial age estimation including, “Our analysis demonstrates that automated age estimation systems cannot be made to work in a reliable way that ensures accuracy, reliability or fairness.”
Unfortunately, the paper is deeply flawed, invalidating the authors’ controversial conclusions.
Firstly, the authors use an old open source 2015 facial age estimation model, which they incorrectly claim is ‘state of the art’. But the training data is not credible. The dataset is relatively small (only 0.5 million faces) and is no longer deemed a reliable set of images and ages for training facial age estimation technology. The faces are mostly of actors captured from the Internet Movie Database (IMD) and the celebrities’ ages are calculated by searching for date of birth from Wikipedia alongside the timestamps of when the photos were probably taken.
A large percentage of male and female actors are known to use makeup and some have undergone cosmetic surgery. Facial photos (and bodies) of celebrities have often been ‘airbrushed’. Unreliable data in means unreliable data out.
The authors also cite many academic papers in support of their claims, but most of these are old (2002 to 2020) in a scientific field that is rapidly advancing and where recent evidence strongly contradicts earlier evidence. The authors even ignore the recent US National Institute of Standards and Technology (NIST) testing that has been providing strong independent evidence that facial age estimation algorithms are now highly accurate. For example, NIST concluded that Yoti facial age estimation is an effective, fair and inclusive way to check age and age ranges with an appropriate threshold.
Additionally, Yoti facial age estimation model was first independently certified by the Age Check Certification Scheme (ACCS) in November 2020 as accurate for under/over 18 age gating. This is one of the key age groups regulators are most concerned about.
Most of the conclusions drawn by the authors are not supported by recent research and independent testing which show that leading facial age estimation models are accurate to perform effective age gating of under/over 13s and 18s. These authors will end up with egg on their faces.