UK to Use Facial Age Estimation for Asylum-Seekers Despite Known Accuracy Flaws

The UK government plans to deploy facial age estimation (FAE) technology at its borders starting next year to estimate the ages of asylum-seekers who lack identity documents. Internal testing reviewed by journalists shows the systems frequently misjudge age, with children sometimes classified as adults—an error that can lead to loss of legal protections and placement in adult detention facilities. The move marks one of the first large-scale uses of AI-based age prediction in an immigration context, raising concerns about accuracy, fairness, and the consequences of automated decisions in high-stakes environments.

Why the UK is turning to facial age estimation for asylum cases

The Home Office has cited the growing number of asylum applications from people without age documentation as a key reason for adopting FAE. When individuals arrive without passports or birth certificates, officials must determine whether they are adults or minors to assign appropriate housing, education, and legal protections. Currently, age assessments rely on interviews, physical exams, and document checks, but these methods are time-consuming and can be inconsistent. FAE promises faster, contactless screening by analyzing facial features such as bone structure, skin texture, and eye spacing to estimate age.

However, the technology is not a neutral tool. It operates on statistical models trained on datasets that may not reflect the diversity of asylum-seekers—especially those from regions with different genetic backgrounds or environmental conditions. Internal evaluations indicate that FAE systems tend to overestimate age in younger individuals and underestimate it in older ones, particularly when skin tone, lighting, or facial expression vary. These biases are not theoretical; they have been observed in controlled tests and could lead to life-altering misclassifications for vulnerable people.

What the internal tests reveal about accuracy and risk

Documents obtained through journalistic investigation detail multiple test runs of FAE systems across different demographic groups. The results show error margins that are far from acceptable for decisions affecting legal status and safety. In some trials, the system incorrectly classified minors as adults in more than one in five cases. Such errors are especially concerning because children misclassified as adults can be denied access to child-specific shelters, educational programs, and legal representation. They may also face harsher treatment during interviews and detention, increasing their risk of harm.

The tests also highlighted how environmental factors degrade performance. Poor lighting, head angles, and even facial expressions influenced outcomes, with some groups—particularly women wearing headscarves—showing higher misclassification rates. These findings underscore that FAE is not a standalone solution but a probabilistic tool with significant limitations. The government acknowledges these risks but argues that FAE can supplement existing methods, not replace them entirely. Still, the lack of transparency about which algorithms are used and how they were trained leaves room for concern about accountability and bias.

Legal and ethical implications of using AI in age assessment

Using AI to determine age in asylum procedures intersects with human rights law, data protection, and ethical AI principles. The UK is bound by international conventions that require age assessments to be conducted in the best interests of the child, with dignity and accuracy. Automated systems that produce unreliable results risk violating these obligations by exposing children to harm through incorrect categorization. Legal experts warn that reliance on flawed AI could lead to legal challenges, especially if decisions are made without human review or appeal mechanisms.

Privacy is another concern. Facial images captured during FAE contain biometric data, which is sensitive and must be handled under strict data protection rules. The Home Office has not publicly disclosed how long images will be stored, who will access them, or whether they could be shared with other agencies. Without clear safeguards, there is potential for misuse or accidental exposure, which could have long-term consequences for individuals seeking protection. Civil liberties groups argue that the use of biometric screening for administrative purposes should be narrowly tailored and subject to independent oversight.

How this fits into a broader trend of AI in immigration enforcement

The UK’s FAE initiative is part of a wider global shift toward using artificial intelligence in border control and immigration processing. Authorities in the United States, European Union, and Australia have deployed or tested AI systems for identity verification, risk assessment, and visa screening. These tools are often promoted as efficiency drivers that reduce wait times and streamline workflows. But critics argue they can embed systemic biases, reduce transparency, and shift responsibility from humans to machines in decisions that affect people’s lives.

Immigration AI systems frequently rely on facial recognition, document analysis, and behavioral profiling—technologies that have faced criticism for high error rates in women and people of color. When such systems are used to make life-altering decisions, the stakes are especially high. The UK’s FAE rollout could set a precedent for other countries considering similar tools, normalizing the use of AI in contexts where errors have direct humanitarian consequences.

What asylum-seekers and advocates should watch next

For asylum-seekers arriving without documents, the introduction of FAE means they may face an additional layer of scrutiny that could influence their legal status. Those who believe they have been misclassified should request a review and insist on human assessment. Legal aid organizations recommend documenting physical appearance, clothing, and any distinguishing features that support a younger age claim. Advocates also advise asking for written explanations of how the AI reached its decision, though such transparency is not guaranteed under current plans.

Civil society groups are calling for independent audits of the FAE systems, public disclosure of error rates by demographic group, and mandatory human oversight in all age-determination decisions. They also urge the government to publish clear policies on data retention, access, and redress mechanisms for those affected by errors. Until these safeguards are in place, the use of FAE remains a high-risk experiment with vulnerable populations.

What businesses and tech providers should consider

For technology providers supplying FAE systems to governments, the UK case highlights the need for rigorous, transparent validation of age estimation models. Vendors must ensure their datasets are representative and regularly updated, and that error rates are disclosed upfront—especially for marginalized groups. Providers should also design systems with human-in-the-loop controls, allowing for manual override and appeal, to mitigate the impact of incorrect classifications.

Businesses in other sectors—such as social media, gaming, or adult content—already use age verification tools that may incorporate similar facial analysis or document checks. They should monitor the UK’s FAE rollout closely, as regulatory scrutiny and public backlash could influence future policy and consumer trust. Companies relying on age estimation should invest in fairness testing, bias audits, and clear user consent mechanisms to avoid similar ethical and legal pitfalls.

What comes after the pilot—and what to demand from policymakers

The UK government has indicated that FAE will be piloted before full deployment, but details about the pilot’s scope, duration, and evaluation criteria remain unclear. Civil society organizations are pushing for mandatory parliamentary review, public consultation, and an independent ethics board to oversee the technology’s use. They also want a moratorium on full-scale deployment until these conditions are met.

Policymakers should consider whether FAE is necessary at all. If the goal is to protect children, alternative approaches—such as trained social workers conducting interviews with interpreters, or improved document verification through international cooperation—may be more reliable and less invasive. Any use of AI in asylum procedures must be demonstrably fair, necessary, and subject to continuous oversight. Without these guarantees, the technology risks becoming a tool of systemic discrimination rather than protection.

The UK’s decision to use facial age estimation for asylum-seekers is a landmark moment in the application of AI to immigration and humanitarian processes. It offers a cautionary tale about deploying unproven technology in contexts where the cost of failure is measured in human lives. As the system moves from testing to implementation, the focus must remain on accuracy, accountability, and the preservation of dignity for some of the world’s most vulnerable people.