Essex Police in the UK is regularly publishing detailed reports on where the technology was deployed, the number of faces scanned, identifications made, interventions conducted and arrests executed.

“Keep people safe and deter crime”

Assistant Chief Constable Andy Pritchard stated: “We’re utilising new technology to keep people safe and deter crime.

“Our Live Facial Recognition technology is used to locate people we want to speak to in connection with ongoing investigations and to manage people with court orders or conditions.

“Criminals cannot think they can walk around our communities without being caught,” Pritchard added.

Statistics

According to publicly available data, Essex Police reported that the the system has demonstrated exceptional accuracy in real-world conditions.

With the findings generated, Essex Police found:

• Across multiple deployments in various cities, Essex Police scanned a total of 383,356 faces while identifying individuals from watchlists containing up to 1,322 names
• The technology generated 61 total alerts, all but one of which were correct, leading to 57 successful interventions and 11 arrests
• The single incorrect alert out of nearly 40,000 scans on that occasion, translates to an error rate of just 0.00026% (or one in 39,401), emphasising the system’s reliability even in complex environments
• The system processed an average of 38,336 faces per deployment, demonstrating its ability to handle high-footfall areas
• Out of 61 total alerts, 57 led to direct police interventions, reflecting a 93% intervention rate
• With 11 arrests made, including sex offenders and individuals suspected of property crime, 18% of all interventions resulted in legal action
• There was an accuracy rate exceeding 99.9997%, reinforcing the technology’s dependability in dynamic policing scenarios

These figures highlight the effectiveness of AI-driven policing tools in safeguarding public spaces while maintaining an exceptionally low rate of false identification.

“Not just about the strength of the algorithm”

Rob Watts, President, Corsight AI explained: “Accuracy is not just about the strength of the algorithm but also a matter of settings.

“Mall of America, for example, has 40 million visitors a year. They want to receive alerts only if an individual is identified with a very high level of certainty, otherwise, security would have to handle a lot of noise.

“On the other hand, police pursuing an armed murderer would want to examine any relevant footage, even if the algorithm is only 90% sure the target is identified.”

Outdoor cameras and facial intelligence

Unlike standard street CCTV cameras owned by municipal authorities, Essex Police deploys police vans equipped with dedicated outdoor cameras and Corsight AI’s facial intelligence software to identify individuals on watchlists.

Since adopting the platform last year, Essex Police has successfully used the technology to enhance public safety and crime prevention. 

“The highest standards”

Tony Porter, Chief Privacy Officer and former UK Surveillance Camera Commissioner, Corsight AI added: “It is a privilege for Corsight AI to contribute to the safety and security of the law-abiding people of Essex.

“Drawing on our vast experience with law enforcement agencies worldwide, Essex Police adhere to the highest standards of fair use and transparency.”