Technology and Science26 February, 20264 min read
Thames Valley Police Wrongly Arrest British-Asian Man After Racist AI Facial-Recognition Misidentification
This article is an automated aggregation of coverage from news outlets. Our editorials offer in-depth analysis curated by the NewsCord team.
Alvi Choudhury, a software engineer from Southampton, was wrongly arrested for a Milton Keynes burglary He was detained and handcuffed at his Southampton home for nearly ten hours
9 outlets told this the same way.
Wrongful facial-ID arrest
Thames Valley Police arrested a man after running CCTV of a Milton Keynes burglary through an automated face-matching search against police records, but the match was wrong and the individual was later released. Brave Search reported that Alvi Choudhury, a 26-year-old software engineer of Bangladeshi origin, was wrongly arrested after Thames Valley Police ran burglary CCTV through a face-matching search against the Police National Database (PND). The Guardian reported that Alvi Choudhury, 26, a software engineer from Southampton, was wrongly arrested in January after Thames Valley Police used automated facial-recognition software that matched him to CCTV footage of a suspect in a £3,000 burglary in Milton Keynes, about 100 miles away.
“Alvi Choudhury, a 26‑year‑old software engineer of Bangladeshi origin, was wrongly arrested after Thames Valley Police ran burglary CCTV through a face‑matching search against the Police National Database (PND).”
The sources contradict each other on Choudhury's background: Brave Search described him as 'of Bangladeshi origin' while The Guardian described him as 'from Southampton'. EasternEye summarised the event with the headline 'UK Police Wrongly Arrest Man in Facial ID Error'.
Alleged wrongful detention
The detained man says officers handcuffed him at home and held him for many hours despite strong alibi evidence and obvious differences between him and the CCTV image. The Guardian reports he 'was handcuffed at home, held in custody for nearly 10 hours and released at 2 a.m.' Attack of the Fanboy describes that he 'was completely shocked, detained for 11 hours despite having a strong alibi and evidence he was at work, and that the CCTV image did not resemble him (different age, skin tone, nose, facial hair and features); an officer reportedly laughed when he pointed this out.'
The Canary similarly says he 'was held for around 11 hours before officers reviewed evidence; the suspect did not resemble him and officers reportedly laughed when releasing him.' The reports disagree on the detention length, citing nearly 10 hours, 11 hours, and around 11 hours.
Biometric identification failures
Reporting highlights technical and data-retention failures that enabled the misidentification. The match reportedly came from an earlier custody photograph retained in police records, and the arrest relied on an algorithmic lead. Brave Search explains: 'The algorithm returned a 2021 mugshot of Choudhury — from an earlier arrest that did not lead to charges — and officers arrested him; he was held about 9–10 hours and later released without charge.'
“The algorithm returned a 2021 mugshot of Choudhury — from an earlier arrest that did not lead to charges — and officers arrested him; he was held about 9–10 hours and later released without charge.”
The Canary notes the match linked 'a custody photo of him taken four years earlier to a suspect in a city he has never visited (about 100 miles from his home).' Attack of the Fanboy and other coverage point to two related failures: 'long-term retention of biometric records for people who were innocent or never charged, and operational overreliance on automated face-matching outputs without sufficient safeguards, training or independent verification' (Brave Search).
Facial-recognition bias reports
Campaigners and multiple reports frame the case as an example of known racial and demographic biases in facial‑recognition technology and call for stronger safeguards. The Canary quotes Liberty's director and others describing 'well-documented racial bias in face-recognition systems — trained predominantly on white faces — making people who are young, women, Asian or Black far more likely to be misidentified,' and says Hart cited a figure that black women were 250 times more likely to be misidentified. LADbible cites published error‑rate figures used to underscore disparity: 'false-positive rates of 5.5% for Black faces and 4.0% for Asian faces versus 0.04% for white faces'.
Attack of the Fanboy notes similar documented higher false‑positive rates. The coverage explicitly links those technical biases to the subject's claim he was targeted 'because they are "a brown person with curly hair"' (LADbible).
Police AI controversy
Thames Valley Police apologised for the distress and said the arrest followed officers’ visual assessment of AI-provided intelligence. The individual has launched legal action and campaigners demand transparency and reform. LADbible records that Thames Valley Police apologized for the distress, saying the arrest was based on officers’ visual assessment following AI-provided intelligence, that the AI did not determine the arrest, later enquiries cleared the individual, and that the force will continue using such tools while aiming to build trust.
“Thames Valley Police apologized for the distress, saying the arrest was based on officers’ visual assessment following AI-provided intelligence, that the AI did not determine the arrest, later enquiries cleared the individual, and that the force will continue using such tools while aiming to build trust.”
Brave Search notes that Choudhury has sued Thames Valley Police and Hampshire Constabulary. The Canary describes wider policy context and concerns, reporting that police AI lead Alex Murray admitted on 24 February that a new £115m national police data centre will produce discriminatory results, and that campaigners demand vendor and system transparency, external audits, removal policies, improved officer training and stronger oversight. Sources use inconsistent name spellings: Brave Search and The Guardian use 'Choudhury' while the Canary uses 'Choudary', a contradiction the sources do not reconcile.
