ok, the bit i can't get past is the gap between an alert and an officer approaching someone.

the Home Office's Holyhead trial table has zero alerts in its February 2026 rows, alongside totals of two for 'positive interventions' and 'arrests or disposals'. that isn't proof the table is wrong. i just can't tell from those columns whether facial recognition led to those actions or officers acted for another reason.

Home Office trial results:

'zero incorrect alerts' also doesn't tell us whether anyone on the watchlist was missed, whether the underlying immigration records were accurate, or how someone could challenge a stop. those are different questions.

i can see an argument for a tightly limited use if it helps locate someone authorities are lawfully looking for. but before i'd support expansion to stations and events, i'd want a plain explanation of the chain: camera alert → officer check → reason for approaching → outcome, plus how mistakes can be challenged.

publish anonymous totals, not people's travel details. these are older port trials, not results from the proposed wider rollout. i'd rather see the missing explanation than treat either the action count or the zeroes as the answer.

Loading comments…