Children’s charities are urging the UK government to abandon plans for AI-powered age estimation at the border, warning the technology could push more child refugees into the adult asylum system.
Their concern centres on racial bias built into the facial-scanning models the Home Office intends to deploy.
According to Humans for Rights Network’s Maddie Harris, the organisation has been working with numerous children that have already been filtered as adults due to the current age assessment process.
Harris referred to the new procedure of using artificial intelligence as a “cheap process” and claimed that it would not make any difference since it is a tool to legitimise the decision made by officials.
The Home Office has released statistics provided by the Helen Bamber Foundation according to which 755 children have been falsely classified as adults while arriving in the UK in 2025. The government itself recognises that even the best artificial intelligence cannot give accurate results while assessing the age of people.
The Home Office selected Cognitec, a German facial-recognition firm, to supply the technology under a contract announced in May.
The spokesperson defended the plans, describing age assessment as “a vital tool in maintaining border security” and framing the AI trial as part of efforts to make the process “more robust and consistent”.
A study conducted by Lighthouse Reports revealed that the system has a tendency to overestimate the age of minors who come from sub-Saharan Africa. This discrimination against young people, according to the rights campaign group Foxglove, is because of the systemic racism contained in the underlying algorithm.
According to Harris, children coming from Sudan and Somalia were particularly vulnerable since their categorisation as adults means that the children will be deprived of protection and become liable to arrest, deportation, and detention among adults due to rising tensions at the asylum accommodation facilities.
Refugee and Migrant Children’s Consortium stated in a separate warning that the stress involved in the journey can age a child’s face and invalidate the assumption on which the technology was based.
Petra Molnar of Migration and Tech Monitor went even further and compared the use of the facial recognition technology to the science of phrenology.
