The British Home Office has signed a three‑year, £322,000 deal with Harlow‑based Akhter Computers Ltd to build AI facial‑recognition tools that estimate migrants' ages at the border. The system, slated for pilot testing in Dover by mid‑2027, is intended to catch adults falsely claiming to be children in order to access asylum benefits.

£322 ,000 contract targets 6,400 disputed child claims

According to the Home Office,more than 6 ,400 migrants who said they were children were assessed for age in the year to March 2026, and almost half were later determined to be adults. The contract will fund development, testing on diverse images, and a trial at a processing centre in Dover, Western Jet Foil, next year.

Human Rights Watch calls the plan “cruel and unconscionable”

Human Rights Watch senior AI researcher Anna Bacciarelli warned that facial‑age estimation has never been used in migration settings and that the technology is unproven. She argued that experimenting with such tools threatens the protections owed to vulnerable children and could lead to wrongful denials of rights.

Current age checks rely on X‑rays, MRIs and officer judgement

At present, trained immigration officers use medical imaging and document verification to gauge age, a process the Home Office says is more costly than AI.. An independent immigration inspector’s report noted frequent misclassifications, underscoring the difficulty of achieving a foolproof method.

Minister Alex Norris says AI will stop “system‑gaming” migrants

Border Security and Asylum Minister Alex Norris defended the initiative, claiming that adult migrants exploiting the system divert vital support from genuine child refugees. he said the AI will help identify, detain and remove those who “game the system” while ensuring protection for those who truly need it.

Who will verify the AI’s accuracy before live deployment?

The contract states that testing will involve images of people of different genders and ethnicities, but the source does not specify an independent audit or external validation before the technology is used in real cases.. This lack of transparent oversight remains a key unanswered question.