Metropolitan Police live facial recognition (NEC NeoFace Live)

entry · p2026-09-07-005 · validated

10 false alerts from the ~ 3,147,436 faces that passed the LFR cameras

tech: met-lfr
passes the definition as: a face-matching neural network fit to labelled face images, run against a watchlist
from: 2024-09-11
region: UK
domain: agency
side: loss
people: 3,147,436 faces scanned (proxy for people) (2025-09-10)
effect: 10 false alerts (2025-09-10) — people wrongly matched to a watchlist; 6 of them were approached and spoken to by officers, none arrested
reversible: no
who holds the switch: owner:Metropolitan Police Service
attribution: named
source kind: primary
the statistic: Met LFR Annual Report: an estimated 3,147,436 faces passed the cameras in the year to 10 Sept 2025 without being asked; 10 false alerts resulted, 6 leading to an officer engagement and none to an arrest, a rate of 1 false alert per 314,743 faces.
source: https://mnd-assets.mynewsdesk.com/image/upload/f_pdf,fl_attachment/pjm94s312mxwbmbj2k0ft3
proposed by: researcher (opus), brief 2026-09-06 second harvest, bot-wall re-run
received: 2026-09-06T10:23:57-04:00
harvest report: impacts/harvest/2026-09-07-botwall-rerun.md

Rulings, in order

2026-09-06T10:24:14-04:00 · validatorpass — quote found on the fetched page (exact match; HTTP 200 application/pdf)
2026-09-06T10:24:14-04:00 · sessionseverity — 3.1 million faces scanned without being asked, ten false alerts, none arrested: reversible, small per person

The proposal as it arrived (JSON)

{
 "id": "p2026-09-07-005",
 "received": "2026-09-06T10:23:57-04:00",
 "proposer": "researcher (opus), brief 2026-09-06 second harvest, bot-wall re-run",
 "kind": "entry",
 "sha": "b0a5373d41281601",
 "data": {
  "kind": "entry",
  "tech": "met-lfr",
  "tech_name": "Metropolitan Police live facial recognition (NEC NeoFace Live)",
  "definition": "a face-matching neural network fit to labelled face images, run against a watchlist",
  "from": "2024-09-11",
  "region": "UK",
  "domain": "agency",
  "side": "loss",
  "people": {
   "n": 3147436,
   "unit": "faces scanned (proxy for people)",
   "date": "2025-09-10"
  },
  "effect": {
   "n": 10,
   "unit": "false alerts",
   "date": "2025-09-10",
   "what": "people wrongly matched to a watchlist; 6 of them were approached and spoken to by officers, none arrested"
  },
  "reversible": false,
  "switch": "owner:Metropolitan Police Service",
  "statistic": "Met LFR Annual Report: an estimated 3,147,436 faces passed the cameras in the year to 10 Sept 2025 without being asked; 10 false alerts resulted, 6 leading to an officer engagement and none to an arrest, a rate of 1 false alert per 314,743 faces.",
  "source": "https://mnd-assets.mynewsdesk.com/image/upload/f_pdf,fl_attachment/pjm94s312mxwbmbj2k0ft3",
  "quote": "10 false alerts from the ~ 3,147,436 faces that passed the LFR cameras",
  "attribution": "named",
  "source_kind": "primary"
 },
 "from_file": "impacts/harvest/2026-09-07-botwall-rerun.md"
}