A Former Police Commander's Perspective on Live Facial Recognition

Live facial recognition (LFR) provides a non-discriminatory way to identify individuals who may be committing crimes at crowded events like Notting Hill carnival. Implemented correctly, LFR can help prevent violence and sexual offenses, while the risk of false positives is relatively low, at one in 33,000. However, concerns about the technology's potential impact on certain groups, such as black people and women, must be carefully considered and addressed through a consolidated statutory framework.

Key Takeaways:

• Live facial recognition technology is used by the Metropolitan police to identify individuals who may be committing crimes at crowded events, with a focus on those on the sex offenders register and those wanted for serious criminal offenses.
• The technology discards live images unless there is a match with a specific police database, and automatically pixelates out anyone else in the shot, minimizing the risk of false positives and protecting the privacy of innocent bystanders.
• The Met's use of LFR has been refined to eradicate past failings that resulted in misidentification of black people and women, with false positives currently running at one in 33,000 for everyone.
• The development and deployment of LFR are overseen by responsible individuals who ensure the technology is used minimally and in a non-intrusive manner, making it less intrusive compared to blanket, no-suspicion "stop and search".
• The former police commander advocates for a consolidated statutory framework that addresses the use of live facial recognition, citing the lack of clear guidelines as a failure of successive governments.
• The Met's LFR system involves officers engaging individuals in conversation to establish their identity and purpose, reducing the risk of unnecessary detention or harassment.

Statistics:

• False positives for live facial recognition are currently running at one in 33,000 for everyone.
• The technology uses specific police databases of facial images for comparison, including those of known sex offenders and individuals wanted for serious crimes.
• The LFR system automatically pixelates out anyone else in the shot, minimizing the risk of false positives and protecting the privacy of innocent bystanders.
• The use of live facial recognition has been refined to eliminate past errors, which previously resulted in the misidentification of black people and women.

Sources:

• Brian Paddick, "I was a cop who was 'soft on drugs' – but I was being a sensible and humane one", The Guardian.
• Brian Paddick, "Let the people decide on live facial recognition", City AM.