AI Risk Wire

Cases › other › LR-2026-0764

British Transport Police live facial recognition trial produces one false identification

Preliminary Why: 1 report: Times of India group · subject to change · 1 independent report

Status: no action reported

Severity moderate
Reported 1 Oct 2026

No source states: loss or records, developer and model, agent tool, could do, deployer, industry.

British Transport Police ran a six-month live facial recognition trial at London railway stations, comparing passing faces with a police watchlist. The trial ran from February to July 2026 across 18 deployments. The system produced one alert against a person on the watchlist. Officers checked it and found the person was not the one sought, so the alert was classed as a false positive, and no arrest resulted directly from the technology. The force then extended the trial by four months and added London Underground stations. Sources differ on the number of faces scanned, with one giving more than 500,000 and another 530,000. Critics raised privacy and proportionality concerns.

What happened

1 · What went wrongThe system produced one watchlist alert, and it was a false identification.
2 · How it was caughtOfficers checked the alert and found the person was not the one sought.

What was done

No source states an action or fix.

Risks

Not yet sorted into a risk

Controls

Not yet read for controls. Controls that mitigate the primary risk:

Evidence

No source states: what the AI was doing, what it reached or did, when it happened, when it was found, how many records were exposed, the direct loss, a count of the other parties affected, any court case, the cause, what set it off, what was done.

Show the full account: 4 items from the primary sources

What happened, in order

  1. February 2026 The six-month trial ran across 18 deployments at railway stations.
  2. 11 February The live facial recognition pilot began.
  3. August The trial was extended for a further four months, adding London Underground stations.

Not dated in the sources:

  • The system produced one alert during the trial, and it was a false identification.
Show every field, with its source and grade

What happened

Earliest source we hold
2026-10-01
D inferred by us from the described behaviour, reason shown
Who caught it
a human reviewer
C one secondary source states it
What went wrong
other
D inferred by us from the described behaviour, reason shown

Impact

Severity
moderate
D inferred by us from the described behaviour, reason shown
Who was affected
More than 500,000 faces scanned across 18 deployments during the February to July 2026 trial, per one source
C one secondary source states it

Notes on gaps

The alert date is not stated, only the trial window.

LR-2026-0764 · version 1 · published 2026-11-01 after automatic checks; not yet checked field by field by a person · JSON