Home Office knew AI age checks for migrant children were flawed — but rolled it out anyway

The Home Office knew an AI tool that will be used to check the age of small boat migrants was flawed but pressed ahead with its roll-out anyway, The Independent can reveal.

The AI-powered technology, which predicts someone’s age based on their facial features, judges teenagers to be adults in error, according to a leaked report the Home Office tried to withhold.

The secret report also found the technology is least accurate when trying to age migrants from countries such as Eritrea and Sudan – which have the highest number of small boat migrants coming to the UK – amid accusations that the technology has “baked-in racial bias”.

It warned that error rates are particularly high for female child migrants from Sub-Saharan Africa, at 4.6 years on average – meaning a 14-year-old girl could be predicted to be an adult.

The report, which was produced in April 2025 by civil servants in charge of the biometrics programme, also warned the tech could be less accurate for people with visible ageing caused by “stress of travel”.

Separately, an audit by Lighthouse Reports of public data on the government’s chosen AI provider shows that its tech misclassified more than a third of 16-year-olds as adults and, in some tests, was shown to give the wrong assessment in 70 per cent of cases.

Scientific advisors to the Home Office have now spoken out for the first time, telling this publication that they felt the government is rushing to adopt the AI tech for political reasons and chose not to consult them to avoid criticism.

Professor Tim Cole, at UCL’s Institute of Child Health, suggested the department is pursuing the technology despite being aware that it is “hideously inaccurate”.

The Home Office announced in May that ‘cutting-edge’ AI tech will be used at the border to curb ‘fake claims by small boat arrivals posing as children’ (Getty)

Ministers announced last month that Facial Age Estimation (FAE) would be used by immigration officers at the border from next year to “crack down on fake claims by small boat arrivals posing as children”.

But, despite knowing about the serious problems, the Home Office said that the “cost-effective” age assessment method had indicated “promising performance and accuracy”.

An investigation by The Independent , in collaboration with Lighthouse Reports and WIRED, has now found the government’s own evaluation of FAE revealed that the technology predicted 17-year-olds to be on average over 18.

In one in 20 cases, the technology said that a 17.5-year-old falls outside the age range of 14 and 22.5. It also performed worse on teenage girls.

Meanwhile, 60 organisations have today sent an open letter to the Home Office calling for a halt to the rollout of FAE, warning that the technology has “baked-in failures and discrimination” and that it is “imprecise at the crucial 16-to-18-year-old boundary”.

Martha Dark, co-executive director at Foxglove, the non-profit organisation that co-ordinated the letter, said children should not be test subjects for “experimental tech that has baked-in inaccuracy and racist bias”, adding: “Errors by these tools could have serious consequences: vulnerable children being forced, alone, into adult detention centres.”

The Facial Age Estimation Performance Test report – which the Home Office refused to disclose through a freedom of information request, but was later leaked to Lighthouse Reports - evaluated seven commercially available algorithms, including those that are ranked top by the National Institute of Standards and Technology (NIST) in the US.

Though the Home Office report did not name which AI algorithms were examined, it has since emerged that the tech used in the roll-out will be provided by German company Cognitec Systems.

Migrants on an inflatable 'small boat' clamber on to The UK Border Force vessel, BSC Intrepid after crossing the channel from northern France on April 27, 2026 in Dover, England. AI facial age estimation technology will be used on migrants at the UK border. (Getty)

The Home Office said any conclusions drawn in the FAE test report did not relate to active procurement of services from any provider.

Separately, the Home Office said in response to a freedom of information request that Cognitec was among seven AI technology suppliers used for internal testing.

Cognitec’s algorithm has been probed by the NIST, with the findings published in May. Analysis of the public data shows that the Cognitec tool misclassified more than a third of 16-year-olds as adults when it was tested on US visa application photos.

This increased to nearly 70 per cent when the AI system was tested on lower quality border crossing photos, data analysis found.

A spokesperson for Cognitec said that the algorithm’s bias was low compared to other algorithms and that they were working to reduce this. They added that the issues with demographic differences affected all vendors and were often related to image quality issues.

The Home Office has not released details about exactly ho…