‘Racist bias’ overestimating ages in Home Office’s facial-recognition software will lead to solo children being housed with adults, says Human Rights Network
Flawed and racialised models that underpin the AI-powered age-detection systems to be introduced by the British government will endanger children, rights groups and children’s charities have warned.
Urging ministers to reverse plans to introduce facial age-estimation technology to screen migrants, critics have warned that black children arriving from conflict zones are at risk of being of thrust into the adult system.
Maddie Harris, of the Humans for Rights Network, said the charity had worked with hundreds of children who were already being wrongly filtered into the adult system during age assessments at borders carried out by immigration officers.
In May, the Helen Bamber Foundation revealed that 755 children in 2025 were incorrectly identified as adults on arrival to the UK
Harris said the introduction of AI was a “cheap process” that would not improve these numbers. “This is not about protecting children – this is about shoring up the decisions that [the authorities] are making,” she said.
The government has admitted that even the best systems can have a 30-month margin of error.
Harris said this is particularly risky for children, especially those from countries such as Sudan and Somalia, who she said were already “adultified” by the system.
“As soon as a child is treated as an adult, they are susceptible to – or exposed, rather – to any of the egregious consequences that the Home Office is applying to adults, such as detention and removal,” she said.
Unaccompanied children would be forced into accommodation alongside adults, exposing them to risk and to the violence threatened by the rise in protests targeting accommodation for asylum seekers, she said.
In May, the government announced a contract to use facial age-estimation technology developed by Cognitec, a German firm that calls itself “the face-recognition company”.
That system showed bias against sub-Saharan Africans, tending to overpredict the ages of minors, according to an investigation by Lighthouse Reports in June.
The UK-based tech monitor Foxglove has said these systems relied on “racist bias” and said the government should not use children to test experimental technology.
The Refugee and Migrant Children’s Consortiumhas raised concerns about the technology’s ability to assess children who have undergone difficult journeys. It noted that trauma can “age” a child’s appearance.
The Home Office plans to use the technology alongside checks by immigration officers. A spokesperson said: “Age assessments are a vital tool in maintaining border security. As part of our work to make this process more robust and consistent, we are exploring cutting-edge AI technology to verify migrants’ ages.”
Petra Molnar, who co-founded the advocacy group Migration and Tech Monitor, said this was the latest in a series of failed attempts across Europe to implement tech and AI into border policy.
She said previous attempts in Europe to implement AI-powered lie-detection systems for screening refugees had proven to be “snake oil” and did not progress beyond pilot schemes.
She likened the use of facial age-estimation technology to “the study of skulls and skull shapes” by colonial-era scientists to categorise different ethnic groups, saying that like the discredited 19th-century pseudoscience of phrenology, these systems similarly relied on categorising physical features to determine age.
“The context really matters here,” said Molnar. “We know that trauma ages you. We know that malnutrition, dehydration, torture – even just the journey itself – has physical impacts on your face, on your body.
“It’s effectively ‘neo-phrenology’ by categorising people based on physical features,” she said. “Relying on these kind of imperfect tools is not the solution.”
Explore more on these topicsReuse this content
