What happened
South Wales Police trialed AFR Locate, live facial recognition with NEC's NeoFace Watch software, at about 50 events between May 2017 and April 2019. Cameras on a van compared faces with watchlists of 400 to 800 people. A police operator reviewed each possible match before officers intervened. A man who said that the cameras had caught him at two Cardiff deployments challenged them. On 4 September 2019 Lord Justice Haddon-Cave and Mr Justice Swift dismissed the claim. They found that the technology had led to arrests or disposals in 37 cases where existing methods could not locate the person. At Queen Street on 21 December 2017 it made 10 possible matches, of which 8 were true. Two arrests followed. The judges wrote that nobody was wrongly arrested. The figure of 37 comes from police evidence. In August 2020 the Court of Appeal allowed an appeal on three grounds, among them that the use was not in accordance with the law. It rejected the proportionality ground.
Sources
-
Court document, by authors with no tie to the system.
High Court of Justice, Queen's Bench Division, Divisional Court sitting at Cardiff (Haddon-Cave LJ and Swift J). R (Bridges) v Chief Constable of South Wales Police & Ors [2019] EWHC 2341 (Admin), judgment of 4 September 2019. Find Case Law copy.
However, the evidence demonstrates that, during the present trial period, this new technology has resulted in arrests or disposals in 37 cases where the individual in question had not been capable of location by existing methods.
Details
AI system
- Workflow
-
Watchlists were built from custody photographs for the purpose of each deployment. Cameras on a marked van fed live images to the software. A police operator checked each possible match before officers intervened.
Follow-up
- Conditions for replication
-
Each deployment is open to the public and limited in time and place. Images of people who are not on a watchlist are processed and discarded at once.
Entry details
- Sector
-
Policing / public protection.
- Purpose
-
NEC built NeoFace Watch to compare faces captured on live CCTV with watchlist images and flag possible matches to an operator.
Not yet checked: model, release, provider, role, configuration, inputs and outputs, autonomy, human oversight, AI contribution, beneficiaries, reach, distribution and access, duration, every measurement question, every money question, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.
Cite this record
Peter H. Vartanian. (2026). AFR Locate led to 37 arrests or disposals in South Wales (AIBD-0835). Artificial Intelligence Benefits Database, edition of 31 October 2026. https://petervartanian.xyz/aibd/records/AIBD-0835.html (archived at https://doi.org/10.7910/DVN/AADSII, version 4.0)