Key facts
- AI facial age-estimation technology planned by the UK government could wrongly classify children as adults.
- Critics warn that racialized models in the AI systems risk placing black children from conflict zones into the adult system.
- Previous Home Office figures revealed 755 children were incorrectly identified as adults in 2025.
- The government acknowledges that even the best age-estimation systems can have a 30-month margin of error.
- The technology, developed by Cognitec, has shown bias against sub-Saharan Africans, tending to overpredict ages.
- Trauma and difficult journeys can physically impact a child's appearance, potentially affecting age assessment.
Charities and rights groups have warned that the UK government's planned introduction of AI-powered facial age-estimation technology for screening migrants could endanger children.
Critics argue that the AI systems, which are based on flawed and racialized models, risk wrongly classifying children, particularly black children arriving from conflict zones, as adults. This could lead to them being processed within the adult immigration system, facing detention and removal.
Maddie Harris of the Humans for Rights Network stated that her charity has already worked with hundreds of children who were incorrectly placed in the adult system during manual age assessments at borders. She expressed concern that the AI technology, described as a "cheap process," would not improve these numbers and is intended to justify existing decisions rather than protect children.
The government has acknowledged that even the most advanced systems can have a margin of error of up to 30 months. Harris highlighted that this is particularly risky for children from countries like Sudan and Somalia, who are already "adultified" by the system. If treated as adults, these children become susceptible to detention and removal, and unaccompanied minors could be housed with adults, increasing their risk of violence.
The technology in question is developed by German firm Cognitec. An investigation by Lighthouse Reports revealed that the system exhibits bias against sub-Saharan Africans, tending to overestimate the ages of minors. The UK tech monitor Foxglove has also criticized the systems for relying on "racist bias" and warned against using children to test experimental technology.
The Refugee and Migrant Children’s Consortium has voiced concerns about the technology's ability to accurately assess children who have endured difficult journeys, noting that trauma can physically age a person's appearance. Petra Molnar, co-founder of the Migration and Tech Monitor, compared the technology to "neo-phrenology," a discredited pseudoscience used by colonial-era scientists to categorize people based on physical features. She emphasized that factors like trauma, malnutrition, and dehydration have visible physical impacts, making reliance on such imperfect tools an inadequate solution.