Age verification is increasingly pervasive online, but its expansion into critical offline applications, such as border control, carries profound implications. The **British government** is set to introduce facial age estimation (FAE) technology to determine the age of asylum seekers, a first-of-its-kind application that could have life-changing consequences. Many asylum seekers arrive in the UK without documentation, and an incorrect age assessment could strip children of vital legal protections, placing them in adult detention centers. An investigation by **WIRED** and **Lighthouse Reports**, in collaboration with The Independent, has unearthed an internal UK government report detailing its FAE system tests. The findings are alarming: these systems regularly mistake children for adults and appear to harbor significant biases. Crucially, the technology performed significantly worse when estimating the ages of Sub-Saharan Africans, who constitute the largest group of migrants arriving in the UK via the English Channel. For female Sub-Saharan Africans, the estimated age was off by an average of 4.6 years, meaning a 13.5-year-old girl could be classified as an 18-year-old adult. This deployment also comes as the **Home Office**, which oversees UK immigration, disbanded a scientific committee designed to advise on age estimation methods while exploring AI solutions. **Tim Cole**, an emeritus professor of medical statistics at **University College London’s Institute of Child Health** and former committee member, described the face scans as “hideously inaccurate” and expressed concern that the committee was shut down before it could highlight these inadequacies. Years of test results from the **US National Institute of Standards and Technology (NIST)** have consistently shown that FAE systems' accuracy varies significantly based on race and photo quality. A **Home Office** spokesperson stated they have “rigorous processes in place” and are working to modernize them, adding that the scientific committee was disbanded due to requiring “different fields of expertise.” While the **Home Office** claims FAE will be an “additional” tool and not “replace or overrule human judgment,” it did not clarify how the technology would be used in real-world scenarios. They affirmed that in cases of uncertainty, individuals would be treated as children until further assessment. ### Expanding Estimates, Persistent Flaws The UK government initially announced its plans in July 2025, with a delayed rollout until 2027. The **Home Office** aims to use this “cutting-edge AI tech” to “crack down on fake claims” and prevent “adults attempting to game the system.” FAE systems function by analyzing facial features, trained on millions of age-labeled faces. In controlled laboratory settings, the best algorithms can predict age within approximately 2.5 years. However, real-world performance is highly variable, influenced by the specific algorithm, gender, demographic details, and image quality. Poor lighting, for instance, can drastically reduce accuracy; some systems have even been tricked by images of video game characters. The leaked **Home Office** report, produced in April 2025 before the technology was purchased, details the testing of seven FAE algorithms against over 2.5 million images. Despite this extensive testing, the “best performing algorithm” still exhibited “substantial deviations” when tested on images of Sub-Saharan Africans. It also tended to over-age 17-year-olds and performed worse on females. Currently, border staff assess age based on physical appearance, interview responses, and demeanor. Since 2010, 40 percent of those undergoing age assessments have been classified as adults. The internal report acknowledged that its findings were based primarily on high-quality images of documented individuals, suggesting that accuracy rates could be even worse in practice, particularly at the “first encounter” where photos are often of lower quality. **NIST** testing corroborates that lower-quality photos typically lead to larger errors. The report concluded that more research was needed on the impact of stress endured by asylum seekers on age estimation results.