UK to Use Facial Recognition to Verify Age of Asylum Seekers Despite Failures

The British government plans to implement facial age estimation (FAE) systems from 2026 to determine the age of asylum seekers arriving without documents. The technology, based on artificial intelligence, scans the face and predicts the individual's age. However, a joint investigation by WIRED, Lighthouse Reports, and The Independent has uncovered an internal government report documenting major failures: the systems frequently confuse children with adults and exhibit biases that disproportionately affect the demographic groups most common among asylum seekers.

This measure represents the first time FAE has been used in an asylum context worldwide. If a minor is mistakenly classified as an adult, they lose legal protections and may be detained in adult detention centers. The report shows the technology is unreliable, but the government plans to use it anyway. This sets a dangerous precedent for other countries that might adopt similar measures.

The measure is part of a global trend toward age verification online (Australia, half of the United States), but now it moves into the physical world with real consequences. Human rights advocates warn that errors could violate the right to asylum and the best interests of the child. Additionally, the internal report indicates that algorithmic biases disproportionately affect certain ethnic groups, which could constitute discrimination.

FAE technology is not reliable for accurately determining age, especially in children. The British government acknowledges the limitations but is proceeding with the plan. Asylum seekers without documents are most vulnerable to errors. Human rights organizations have already expressed opposition. This case could set a precedent for the use of AI at borders.

“It is the first time facial age estimation has been used in an asylum context, and the risks are enormous,” the investigation states.

Historical context of age verification

Age verification is not new. For decades, physical documents such as passports or driver's licenses have been used to confirm age in contexts like purchasing alcohol or accessing adult content. However, the digital age has driven automated methods. In 2023, Australia passed a law banning minors under 16 from accessing social media, leading platforms like Meta to implement age estimation systems. In the United States, more than half of states have introduced age verification laws for pornographic sites, according to 404 Media. These laws have sparked controversy over their impact on privacy and free speech. The British case marks a turning point by applying FAE in a physical and high-risk context, such as determining the age of asylum seekers.

Details of the internal report and concrete data

The government report, obtained by WIRED and Lighthouse Reports, evaluated multiple FAE systems from providers such as Yoti and Herta. According to the analysis, the systems showed significant error rates when estimating the age of children, especially those aged 12 to 17. In some cases, the false positive rate (classifying a minor as an adult) exceeded 20%. Additionally, algorithmic biases disproportionately affected people with darker skin and certain ethnic backgrounds, who make up the majority of asylum seekers in 2025 according to Home Office data. The report also notes that accuracy decreases under varying lighting conditions or when the subject displays non-neutral facial expressions. Despite these shortcomings, the British government has contracted Yoti to implement the system in asylum processing centers, with no plans for additional independent testing.

Impact on asylum seekers and human rights

The consequences of misclassification are severe. A minor classified as an adult loses the right to be housed in specialized centers for children, to receive legal guardians, and to have their case prioritized. They may be detained in adult facilities, where there is a higher risk of abuse and trauma. Organizations such as Amnesty International and the Refugee Council have expressed concern, noting that the technology does not meet international asylum law standards, which require that in case of doubt, the applicant is presumed to be a minor. Furthermore, ethnic bias could violate Article 14 of the European Convention on Human Rights, which prohibits discrimination. The case could reach the courts if FAE is shown to be used in a discriminatory manner.

Comparison with other uses of AI at borders

This is not the first attempt to use AI at borders. In 2020, the U.S. Immigration and Customs Enforcement (ICE) artificial intelligence system was criticized for racial bias in fraud detection. In the European Union, the Eurodac system uses fingerprints to identify asylum seekers but has been questioned for its accuracy. However, FAE represents a qualitative leap: unlike fingerprints or DNA, AI-estimated age is not a precise biometric measure but a probabilistic prediction. This makes it inherently less reliable. Moreover, while other systems focus on identification, FAE seeks to determine a subjective attribute like age, which can vary based on genetic and environmental factors. Computer vision experts, such as Professor Sandra Wachter of Oxford, warn that “facial age estimation is not mature enough for life-or-death decisions.”

Implications for the market and technology

The British case could drive adoption of FAE in other countries, such as Australia, which is already exploring its use at borders, or the EU's entry/exit system. Companies like Yoti, which already provides age verification systems for online platforms, see a new market in the government sector. However, public and legal scrutiny could slow this expansion. If systems are proven defective, companies could face lawsuits for damages. Additionally, the controversy could affect trust in online age verification, where similar technology is used. On the other hand, regulatory pressure could accelerate the development of more accurate and equitable systems, though for now evidence shows the technology is not ready.

What readers should know

• FAE technology is not reliable for accurately determining age, especially in children.
• The British government acknowledges the limitations but is proceeding with the plan.
• Asylum seekers without documents are most vulnerable to errors.
• Human rights organizations have already expressed opposition.
• This case could set a precedent for the use of AI at borders.
• Documented ethnic bias could violate anti-discrimination laws.
• The measure is part of a global trend, but with unique risks in the asylum context.

“It is the first time facial age estimation has been used in an asylum context, and the risks are enormous,” the investigation states.