Ethnic minorities four times more likely to face South Wales Police facial recognition scans

Nov 16, 2026
They made up 24% of people checked with officer-run facial recognition in 2025 but 7% of the population, Freedom of Information data shows.

Ethnic minority people were four times more likely than white people to be selected for officer-initiated facial recognition checks by South Wales Police in 2025, according to data obtained under the Freedom of Information Act by The Guardian. People from ethnic minorities made up 24% of those scanned, despite being 7% of the area's population and 11% of people arrested.

The checks involve officers photographing a person's face on a smartphone and running it through the Neoface algorithm against custody mugshots and other face databases. South Wales Police, which covers Cardiff and Swansea, is one of the first UK forces to use the technology routinely, to identify suspects, wanted people and missing persons who decline to give their details, as well as people found unconscious or dead. The Metropolitan Police is trialling a similar system, and it has been tested in Liverpool.

Civil liberties campaigners called the figures "alarming". The force said they raised "concerns about disproportionality" and that it was reviewing its approach. Last week the police watchdog warned that innocent people, especially from ethnic minorities, risked wrongful arrest if police treated inaccurate retrospective facial recognition results as evidence.

We hope you enjoyed this article

Consider subscribing to one of our newsletters like Daily AI Brief.

Also, consider following us on social media:

Free newsletter

Daily AI Brief

Daily report covering major AI developments and industry news, with both top stories and complete market updates

Whitepaper

Tensordyne Napier: What If One Rack Could Do the Work of Nine?

Tensordyne

This Tensordyne whitepaper presents Napier, an inference-focused AI processor and rack-scale system based on the company’s TDN Math logarithmic number system. It examines infrastructure requirements for large mixture-of-experts and agentic models, compares major inference architecture approaches, and details the TDN AIP processor, TDN72 pod, TDN Link fabric, and Napier Ultra configuration. The paper reports simulation-based performance, cost, and accuracy-validation results, including Tensordyne’s projected comparison of one Napier rack with a nine-rack Nvidia Rubin plus Groq deployment; the chip is reported as taped out and in fabrication.

Read more
Free, six days a week

Daily AI Brief: the AI news that matters, in your inbox.