From phrenology to performance data, UK immigration age estimation debate intensifies

Advocacy groups are attacking the use of biometric facial age estimation as evidence in evaluations of child status among immigrants to the UK. Some of the criticisms reflect generalized scepticism with the technology, but others make more specific points about the challenges of applying normal expectations about how people age, whether human or algorithmic, to people in extreme circumstances.

There were 755 children incorrectly identified as adults by the UK without the benefit of facial age estimation in 2025, according to Home Office figures uncovered by the Helen Bamber Foundation. Home Office processes 111,000 asylum claims a year, with 7 percent undergoing age assessments. The government found 17 percent of those were incorrectly identified as children between July and December of 2025.

Home Office awarded a contract for 322,000 pounds (roughly US$431,000) in May that makes Cognitec the FAE supplier for the UK immigration system.

When the decision was announced, Home Office said that “(d)ecisions informed by a wider range of information, made with greater confidence, will help ensure that vulnerable children get the support and protections they need.”

But critics argue that the technology is not yet mature enough to handle a task that involves the very demographic groups it struggles the most with.

An investigation co-published in June by Lighthouse and Wired shows that Home Office figures peg the mean absolute error (MAE) for girls from Sub-Saharan Africa is 4.6 years.

Cognitec submitted an algorithm to the Face Analysis Technology Evaluation (FATE) for age estimation run by the U.S. National Institute of Standards and Technology (NIST) in April, which had higher error rates for people with darker skin, and misclassified a majority of 16 year-olds as adults using border images. Across the four tested datasets, Cognitec’s algorithms consistently classified more than half of West African 16 year-olds as over 18.

A representative of Humans for Rights Network told The Guardian that many children are “adultified” by the current system, but the introduction of FAE is motivated by a desire to back up those decisions. Foxglove called the technology experimental and said it exhibits “racist bias.”

A Migration and Tech Monitor co-founder compared age estimation to phrenology, dubiously claiming that the pseudoscience which purported to measure mental characteristics through physical ones “similarly relied on categorising physical features to determine age.”

A more pertinent point is raised by a spokesperson for the Refugee and Migrant Children’s Consortium, who points out that appearance can change due to trauma, which could reduce the effectiveness of age estimation.

In a statement emailed to Biometric Update, Cognitec Marketing Manager Elke Oberg noted that the demographic differences noted are found in all algorithms assessed for NIST’s FATE AEV.

“The reasons for bias are extremely complex and often related to image quality issues,” Oberg says. “Different testing methods and test images yield varied results. Please note that the bias of Cognitec algorithms is low compared to other algorithms of similar overall accuracy, and be assured that we are diligently and continuously working on reducing bias by developing specific testing methodologies, designing loss functions in our network training, and by diversifying the training and testing data.”

Article Topics

age verification  |  biometric bias  |  Cognitec  |  face biometrics  |  facial age estimation (FAE)  |  Home Office Biometrics (HOB)  |  immigration  |  NIST