Blog
Original writing on artificial intelligence in policing and criminal justice: weekly briefings on what changed, and longer analysis of the systems, rulings and evidence behind the record this site maintains.
The backlash against licence plate cameras did not start with the cameras. It started with the search logs.
In eight days, Florida revoked permits for licence plate readers on state highways, a bipartisan bill was introduced to strip federal funding from them, and a Senate investigation widened to three more suppliers. The cameras have been going up for years without much argument. What changed is that the record of who searched them, and why, became readable by the public.
Read the briefing →Dubai Police announced 36 robots. None of them have bodies.
Three separate stories reached the archive today about Dubai Police deploying 36 robots, one of them under the word robocops. The force's own announcement describes robotic process automation: software that moves data between systems. The gap between those two descriptions is the entire story.
The system placed 25 people at a protest. Prison records placed them in jail.
An investigation checked 205 of the 2,873 people Delhi Police told India's Supreme Court its facial recognition system had identified at the Jantar Mantar protests. Prison records placed at least 25 of them in custody at the moment they were flagged. What makes the finding unusual is not the number. It is that anyone was able to check at all.
Facial recognition gets the scrutiny. The national police data service is the bigger change.
Live facial recognition attracts the arguments, but the National Data Integration and Exploitation Service would change what every other police AI system in Britain is able to see. It is being designed now, valued at up to £250m over five years, with no supplier announced and several basic questions unanswered.
The best public record of UK facial recognition is the one forces publish themselves
Essex, Surrey, Sussex and Merseyside all now publish material about their own facial recognition deployments. These pages are the strongest public evidence available on how the technology is being used. No two forces publish the same things, and on one force's page the figures do not reconcile with each other.
When AI writes the police report, who owns the record?
AI-assisted police reports promise to save officers time. But once machine-generated text becomes part of an evidential record, transparency, audit trails and human verification become much harder questions.
Flock’s new AI changes the question from plate reading to relationship mapping
Flock’s network is no longer just about finding a plate. New AI can infer vehicle patterns and associations, making governance as important as the camera itself.
Retrospective facial recognition runs at far greater scale than live, with less scrutiny
25,000 searches a month against 19 million custody images, a confirmed algorithm bias, and a regulator finding compliance worse than for live deployment. The quiet version of facial recognition is the bigger one.
Sixteen police AI systems that were switched off, and why
Most coverage of police AI is about adoption. This is the other half: sixteen documented shutdowns, grouped into four recognisable patterns.
Regulators and inspectorates pull in opposite directions on facial recognition
The ICO found police facial recognition governance needs urgent work. Two days later, Scotland's chief inspector urged faster adoption. Both landed in the same week.