London station facial recognition trial: 500,000 scans, one false alert, no arrests from it
A six-month British Transport Police trial of live facial recognition in London stations cost over £320,000 and produced one watchlist alert, which was wrong. BTP says other arrests were made during deployments and has extended the trial.
1 / 3
The story, neutrally told
A British Transport Police (BTP) trial of live facial recognition at London railway stations, from February to July, scanned more than half a million faces and produced one watchlist alert, which was a false positive. The GuardianLC “led to one false match against a watchlist of suspects and no arrests” Read at The Guardian ↗ GB NewsR “the entire exercise generated just one watchlist alert - which proved to be a false positive” Read at GB News ↗ Freedom of information documents obtained by Liberty Investigates and shared with the Guardian put the cost of equipment hire and staffing for 18 deployments at £320,786, plus almost 100 police hours. The GuardianLC “equipment hire and police staffing for the 18 deployments across the trial cost £320,786” Read at The Guardian ↗ GB NewsR “Equipment hire and officer staffing consumed £320,786 and nearly 100 hours of police time” Read at GB News ↗ The technology aimed to help catch offenders and people breaching court orders. The GuardianLC “which aimed to help catch offenders and people breaching court orders” Read at The Guardian ↗
BTP announced last month that it would extend the trial by four months and add London Underground stations. The GuardianLC “Last month BTP announced it was extending the trial for a further four months” Read at The Guardian ↗ SlashdotN “BTP has extended the pilot through November and expanded it to selected Underground stations” Read at Slashdot ↗ BTP says that since the extension there have been three confirmed alerts of people found to be complying with sexual harm prevention orders or other court conditions. The GuardianLC “three positive confirmed alerts of people who were subsequently confirmed to be complying with sexual harm prevention orders” Read at The Guardian ↗ A BTP spokesperson said officers made a number of associated arrests during deployments, but these did not result directly from an LFR alert and so are excluded from performance data. The GuardianLC “As these arrests did not result directly from an LFR alert, they are not included within LFR performance data.” Read at The Guardian ↗
Fraser Sampson, a former biometrics and surveillance camera commissioner, said police must show proportionality and that the trial does not appear to have been very fruitful; the Guardian notes he is a non-executive director of Facewatch. The GuardianLC “In that respect the trial doesn’t appear to have been very fruitful.” Read at The Guardian ↗ Labour MP Bell Ribeiro-Addy called for the rollout to be suspended until there is a proper legal framework, while the Met and the mayor's office are installing fixed cameras in the West End. The GuardianLC “The government should be suspending their rollout until there are stronger safeguards in place”“fixed LFR cameras would be installed in the capital’s West End” Read at The Guardian ↗
Every sentence links to the reporting it rests on.
Left1 outlet
- Framing
- Investigative, privacy and proportionality angle built on the FOI findings, with expert and MP criticism and wider rollout context.
- Emphasis
- Cost, oversight, human rights concerns, expanding deployments.
- Leaves out or plays down
- Little on supporters' case beyond BTP and TfL statements.
- Charged language
- “surveillance technology”
- For example
-
“Live facial recognition technologies are a serious invasion of our privacy.” — The Guardian
Centre1 outlet
- Framing
- Short aggregation summarising the Guardian report around the headline numbers.
- Emphasis
- Scan count, cost, zero arrests, extension.
- Leaves out or plays down
- Text is partial; BTP's note on other arrests is not visible.
- For example
-
“produced just one alert with no arrests directly resulting from the technology.” — Slashdot
Right1 outlet
- Framing
- Value-for-money angle: taxpayer money spent for a wrong result.
- Emphasis
- Cost and lack of arrests.
- Leaves out or plays down
- Visible text does not include BTP's statement on associated arrests or the critics.
- Charged language
- “underwhelming results”“surveillance technology”
- For example
-
“Despite these underwhelming results, BTP announced last month that it would extend the pilot” — GB News
What every side reports
- The trial ran six months across London stations and scanned over 500,000 faces.
- It cost about £320,000.
- It produced a single watchlist alert, which was a false positive.
- BTP has extended the trial by four months, adding Underground stations.
Where accounts differ
-
Whether the trial produced arrests
- Left
- No arrests from the trial's alert, but BTP says other arrests were made during deployments that are not counted as LFR results.
- Centre
- No arrests directly resulting from the technology.
- Right
- Says the trial failed to produce a single arrest, without the BTP caveat in the visible text.
-
The six-month BTP live facial recognition trial cost about £320,000 and produced one watchlist alert, which was a false positive.
Supported- Supports 2
- The Guardian, GB News
- Reports 1
- Slashdot
-
BTP says officers made other arrests during deployments that did not result directly from LFR alerts.
Reported- Disputes 1
- GB News
- Reports 2
- The Guardian, Slashdot
British Transport Police organisation
Says the pilot aims to identify wanted offenders and risks, that it has refined deployment and watchlists, and that other arrests occurred outside LFR data.
“watchlist growth has been deliberate, controlled and aligned to BTP policy” — The Guardian
Left1 article
-
Trial of live facial recognition in London stations leads to a false positive and no arrests
Critical FOI-based investigation questioning the trial's value and proportionality.
Centre1 article
-
500K Facial Recognition Scans at UK Stations Yield Zero Arrests, One False Positive
Neutral Condensed relay of the Guardian report with numbers in the headline.
Right1 article
-
GB NewsR ·
Facial recognition trial at railway stations cost £320k and led to just one alert... which was wrong
Critical Presents the trial as costly and ineffective.
- 29 Sep 10:48 First The GuardianLC Trial of live facial recognition in London stations leads to a false positive and no arrests
- 29 Sep 20:00 +9h 12m SlashdotN 500K Facial Recognition Scans at UK Stations Yield Zero Arrests, One False Positive
- 30 Sep 06:18 +19h 30m GB NewsR Facial recognition trial at railway stations cost £320k and led to just one alert... which was wrong
Times are when each article was published, or when we first saw it if the outlet gave no time.