School of Law academic advises Parliament on facial age estimation (FAE) technology

Professor Subhajit Basu’s Parliamentary Office of Science and Technology (POST) rapid-response article on facial age estimation has been published.

POST rapid response briefings are impartial, evidence-based summaries for Parliament to help it in its work.

Professor Basu was approached directly by the POST team in May 2026, and he produced a POST rapid-response article based on his expertise in law and technology. His time was supported by the Research and Innovation Policy Support Fund.

What is FAE and where is it used?

Facial age estimation uses AI to analyse a person’s face. An uploaded image is converted into a numerical representation of the face, then a model trained on faces of known ages uses the numerical representation to give an estimated age or age range. Many people are already encountering FAE without realising it, as it is being used by platforms including Instagram, Sony PlayStation and OnlyFans, and has been trialled at supermarket self-checkouts for age-restricted purchases.

How accurate is it, and how is it regulated?

The ‘best’ systems still have a ‘mean absolute error’ of around 2.5 years at the critical 16 to 18 boundary according to Home Office guidance, with higher error rates for female faces. Oversight remains fragmented, and regulation spans data protection, human rights and equality law, alongside oversight bodies such as biometrics commissioners.

Immigration and ethical concerns

The Home Office plans to test FAE in 2026, with possible use along the UK border from 2027. Concerns are that if a child is wrongly identified as an adult they could lose safeguarding protections, and if an adult is wrongly identified as a child they could pose safeguarding risks to other children placed with them.

Some say FAE is less intrusive than technologies used to identify a person and retain their image, while others say that it may ‘give rise to considerable concern regarding data privacy and security.’ There is also a risk of bias, as FAE performs differently by ethnicity, skin tone, gender, place of birth, and quality of input image.

Professor Basu poses the following questions for Parliament:

What level of verified accuracy for FAE is sufficient?
Should stricter standards apply in high-risk situations like immigration?
Should independent audits be mandatory and published?
What oversight is needed as the technology evolves?

Professor Basu says:

“Facial age estimation is often presented as a less intrusive alternative to facial recognition because it estimates age rather than seeks to identify an individual. But that distinction should not distract us from the consequences of getting it wrong.”

In areas such as immigration and safeguarding, an inaccurate assessment could determine whether a person is treated as a child or an adult, with potentially serious implications for their rights, safety and access to protection.

“The key question for Parliament is therefore not simply whether the technology can be used, but what level of accuracy, independent scrutiny and legal accountability should be required before it is relied upon in decisions of such importance.”

Professor Basu’s engagement with this POST briefing reflects the University’s support for policy-facing work, and demonstrates the School of Law’s strong links with Parliament.