A WIRED investigation published in June 2026 has revealed the full scope of Avon and Somerset Police’s predictive analytics program in Bristol, England — a system that assigned algorithmic risk scores to thousands of adults and children, operated largely out of public view, and saw at least two of its models quietly abandoned after officials deemed their results untrustworthy.
At the center of the program is the Think Family Database, launched in 2016 by Bristol City Council and Avon and Somerset Police. The database holds records on nearly half a million people and draws on sensitive data including police intelligence reports, housing status, mental health records, teenage pregnancies, and free school meal enrollment. Police data scientists used this information to build at least 23 separate machine-learning models designed to predict outcomes such as the likelihood of someone committing burglary, failing to appear in court, going missing, or becoming a victim of domestic abuse.
A separate tool, the Offender Management App, was designed to hold data on around 300,000 people in the region. One senior officer described it as a “league table” of the area’s most dangerous criminals. John Pegram, a local police accountability group leader in Bristol, says he did not learn about the app until 2023 — years after it was created. When he filed a data request in early 2024, police initially refused to respond. After he hired solicitors, police confirmed he was on the app but provided no further detail.
WIRED, working with Liberty Investigates, the Bristol Cable, and Lighthouse Reports, obtained hundreds of pages of documents through public records requests. The records show government inspectors and independent reviewers raised concerns about a lack of transparency and warned the systems could undermine public trust. An independent analyst who reviewed more than 36,000 model performance scores disclosed by police found that some appeared to show “genuinely poor predictive performance.”
The investigation suggests that residents across Bristol — and potentially beyond — may have been scored by algorithms they had no knowledge of, with no clear understanding of how those scores could affect their interactions with authorities.
Source: WIRED