The UK Will Scan Asylum-Seekers’ Faces for Age Checks—Despite Knowing the Tech Is Flawed
The UK Will Scan Asylum-Seekers’ Faces for Age Checks—Despite Knowing the Tech Is Flawed
By Decode Today News
The British government is moving forward with plans to implement facial age estimation (FAE) technology to determine the age of asylum seekers, a decision coming under intense scrutiny after an internal government report revealed serious flaws and biases within the systems. Set to be rolled out by 2027, this initiative marks a pioneering use of AI-driven age prediction in a high-stakes, real-world immigration context, despite clear evidence that the technology regularly misidentifies children as adults and exhibits significant demographic inaccuracies.
An investigation by WIRED and Lighthouse Reports, in collaboration with The Independent, uncovered a leaked UK Home Office document detailing its tests of FAE algorithms. The findings painted a stark picture: the systems performed significantly worse when estimating the ages of Sub-Saharan Africans, who represent the largest group of migrants subject to age assessments in the UK. For female asylum seekers from Sub-Saharan Africa, the technology’s age prediction was off by an average of 4.6 years, meaning a 13.5-year-old girl could be incorrectly assessed as an 18-year-old adult. Such misclassifications carry profound consequences, as children mistakenly deemed adults can be deprived of legal protections and placed in adult-only detention centers.
The Home Office's internal report, compiled in April 2025 before the government acquired its facial scanning technology, explicitly detailed how even the "best performing algorithm" showed "substantial deviations" when tested on images of Sub-Saharan Africans. It also tended to predict that a 17-year-old would be over 18 and performed worse on females overall. This means the UK government was aware of the technology's shortcomings before committing to its deployment.
Critics have voiced strong concerns. Tim Cole, an emeritus professor of medical statistics at University College London’s Institute of Child Health and a former member of a scientific committee advising the Home Office, described the face scans as "hideously inaccurate." He revealed that his committee, designed to offer guidance on broader age estimation methods, was disbanded by the Home Office while it was exploring AI, preventing them from highlighting the inadequacies of FAE. "We were keen to highlight the inadequacies of facial age estimation, but this opportunity was not presented to us, and then the committee was shut down," Cole stated.
A Pattern of Bias and Uncertainty in AI Tech
The issues identified in the UK's internal report align with years of independent research. The US National Institute of Standards and Technology (NIST) has consistently shown that FAE systems’ accuracy is often dependent on the race and gender of the person being analyzed, alongside the quality of the photographic input. These systems, trained on millions of age-labeled faces, can predict age within about 2.5 years in controlled environments. However, real-world conditions, especially poor-quality images or those depicting individuals under stress, drastically reduce their reliability.
For asylum seekers arriving in the UK, often after perilous and traumatic journeys across the English Channel in small boats, the conditions for accurate facial scanning are far from ideal. The Home Office's own leaked report noted that photos taken at initial encounters were "routinely worse" than subsequent images. This poor quality, combined with the "temporary aging" effect of trauma and travel stress, appeared to significantly impact the algorithms' performance, a factor the report concluded needed more study.
While the Home Office stated that FAE technology would serve as an "additional" tool for border officers and not "replace or overrule human judgment," questions remain unanswered about its practical application. The department has not clarified how the technology will be integrated into real-world environments, nor how border staff, who already work under pressure, will interact with these potentially flawed AI estimates. They did, however, state that "in cases of uncertainty, individuals will always be treated as children until a further assessment is conducted."
Global Implications and Public Outcry
The UK's move comes amidst a global trend where governments, including a potential second Trump administration, are increasingly adopting anti-migrant policies and investing heavily in surveillance technology. This technology is often deployed against vulnerable populations who frequently lack understanding of how these systems work or how to challenge their outcomes. The use of FAE by the UK could set a significant precedent for other nations considering similar applications.
The Home Office initially announced its plans in July 2025, aiming to use "cutting-edge AI tech" to "crack down on fake claims" and prevent "adults attempting to game the system." Despite the internal report's findings, and a delay in rollout to 2027, the department spent over $400,000 in May on face-scanning technology from German company Cognitec. While it's unclear if Cognitec's system was the "best performing algorithm" in the internal report, an analysis of Cognitec's public data by WIRED and Lighthouse Reports showed its system misclassified 16-year-olds as 18 or older twice as often when using lower-quality border photos compared to high-quality visa photos. A Lighthouse Reports audit of NIST data further revealed that 16-year-olds from West Africa were more likely to be misclassified than their Eastern European counterparts.
Cognitec, while unable to comment on specific work with the Home Office, acknowledged that "demographic differences" and "bias" apply to all face scanning algorithms. A company spokesperson stated, "The reasons for bias are extremely complex and often related to image quality issues," and assured that they are "diligently and continuously working on reducing bias."
Human rights advocates have strongly condemned the plans. Martha Dark, co-executive director of rights group Foxglove, argued, "Children seeking asylum have often suffered unimaginable trauma. They should not be the test subjects for experimental tech that has baked-in inaccuracy and racist bias." Foxglove, alongside 61 other organizations, has sent an open letter to the UK government, urging the Home Office to abandon its plans.
The Unsettling Future of AI in Immigration
The introduction of AI into sensitive immigration processes highlights a complex challenge. While the Home Office's human-led age assessments have historically faced their own problems, including "poor" record keeping, "perfunctory" visual assessments, and a lack of specific training for staff until 2023, replacing these with flawed AI presents new and equally concerning risks. The department's guidance suggests AI will allow immigration officers to "test their judgment against the technology’s estimate," yet the very basis of that technology is steeped in documented inaccuracies and biases, particularly affecting the most vulnerable.
The Home Office has yet to provide detailed answers to critical questions posed by the investigating journalists, including how stretched border officers will truly interact with these AI estimates, whether specific training will address the systems' known weaknesses, and if standards will be set for photograph quality and conditions. The reliance on experimental tech for life-altering decisions, especially when its flaws are clearly understood, raises serious ethical questions about the balance between technological advancement and human rights.
As the global digital culture increasingly integrates AI into everyday life, the case of the UK's facial age estimation for asylum seekers serves as a potent reminder of the profound human cost when technology, however promising its intent, is deployed without adequately addressing its inherent biases and limitations. For asylum seekers, the promise of modernization could translate into devastating misjudgments, impacting their legal status, safety, and future.
