The introduction of AI-powered age-detection systems by the British government has sparked concerns among rights groups and children's charities, who fear it will further endanger child refugees. This technology, intended to screen migrants, carries inherent flaws and racial biases, which critics argue will disproportionately affect black children from conflict zones, pushing them into the adult system.
Maddie Harris, from the Humans for Rights Network, highlights the charity's experience with hundreds of children who have already been misclassified as adults during border assessments. The Helen Bamber Foundation's revelation that 755 children were wrongly identified as adults in 2025, based on Home Office data, underscores the urgency of this issue.
Harris believes the government's decision to adopt AI is more about reinforcing existing decisions than protecting children. Even the best AI systems admit to a 30-month margin of error, a risk that is particularly concerning for children, especially those from countries like Sudan and Somalia, who are already at a disadvantage due to systemic 'adultification'.
The consequences of misclassification are severe. Unaccompanied children could be placed in adult accommodations, exposing them to potential violence and the rising protests targeting asylum seeker housing. This raises ethical questions about the government's commitment to child protection and its willingness to experiment with technology that has the potential to cause harm.
The government's contract with Cognitec, a German firm specializing in facial recognition, has also come under scrutiny. An investigation by Lighthouse Reports found that Cognitec's system exhibited bias against sub-Saharan Africans, overpredicting the ages of minors. UK-based tech monitor Foxglove has criticized these systems for relying on 'racist bias' and urged the government not to use children as test subjects for experimental technology.
The Refugee and Migrant Children's Consortium has expressed concerns about the technology's ability to accurately assess children who have endured traumatic journeys. Trauma, they argue, can physically 'age' a child, making it even more challenging for AI systems to provide accurate assessments.
The Home Office defends its plans, stating that age assessments are crucial for border security and that AI technology will enhance the robustness and consistency of the process. However, Petra Molnar, co-founder of the Migration and Tech Monitor, sees this as another failed attempt by European governments to integrate technology and AI into border policy.
Molnar draws parallels between facial age-estimation technology and the discredited pseudoscience of phrenology, where physical features were used to categorize and judge individuals. She emphasizes the importance of context, noting that trauma, malnutrition, and the journey itself can have physical impacts that AI systems may misinterpret. Relying on such imperfect tools, she argues, is not a viable solution.
In conclusion, the proposed use of AI-powered age-detection systems raises serious ethical and practical concerns. As we navigate the complexities of border security and refugee protection, it is essential to prioritize human rights and ensure that technology serves to enhance, rather than undermine, our commitment to vulnerable populations.
