Charities warn over UK AI age checks for young asylum seekers

TL;DR:

• A coalition of more than 100 refugee children’s organisations says Home Office plans to use AI facial age estimation on disputed-age asylum seekers could push more children into adult prisons and detention centres.
• The warning follows a £322,000 three-year contract awarded to Akhter Computers Ltd, with technology that analyses photographs of small-boat arrivals at Dover and a national rollout targeted for 2027.
• Campaigners are not calling for an outright ban, but want AI used only in an advisory role alongside human social-work assessments, with appeal rights and safeguards.

The Home Office’s decision to deploy facial age estimation on young asylum seekers has drawn pushback from the Refugee and Migrant Children’s Consortium, whose forthcoming report argues the technology risks hard-coding existing errors rather than fixing them. The department frames the move as a way to stop adults “gaming the system” by posing as children.

Where the concern lies

The dispute turns on accuracy and consequence. Home Office data already shows young asylum seekers are more than twice as likely to be assessed as children by social workers than by immigration officers at the border, with over two-thirds judged to be minors. Campaigners argue that trauma, malnutrition and exhausting journeys alter appearance in ways an algorithm cannot read, and that biased training data and poor image quality compound the risk.

Border security minister Alex Norris said the rollout would identify those who “game the system” while protecting genuine minors, with final decisions still resting with immigration officers. The consortium’s co-chair Kamena Dorling countered that AI “faces the same problems with bias and inaccuracy as human decision-making”. The report’s central recommendation is that the tool inform, not determine, outcomes — backed by access to an appropriate adult, legal advice and a right to challenge.

Looking forward

For UK readers, this is an early test of how the government deploys AI in high-stakes public services where errors carry severe human cost. The technology faces further testing before its 2027 rollout, leaving a window for the safeguards campaigners want. The wider question is whether automation introduced to cut disputed-age caseloads ends up lending false certainty to decisions that remain difficult to get right — a tension that will recur as AI spreads across Home Office and welfare functions.