Live facial recognition resumes for London Met as law enforcement seeks wider use
The appeal of biometrics and facial recognition in particular to law enforcement around the world is getting ever clearer. A host of recent developments shows how police forces may be starting to use more of the technology, as some restrictions are repealed and others potentially ignored. Facial recognition is also making its way the ‘management’ of homeless people.
Virginia lawmakers prepare to repeal checks on police use
Some of the most stringent restrictions on police use of facial recognition in the U.S. may be repealed as some Virginia lawmakers see their July 2021 introduction as a temporary measure, reports the Virginia Mercury. The strict measures were brought it when it was revealed by The Virginian-Pilot that police were using the tech with little oversight and no public disclosure.
A replacement bill would make tools available to police with reasonable suspicion of a crime being committed or planned, but legislation would not allow mass surveillance in the state, according to senators. It could also be used to identify victims or witnesses of crimes. Any software used would have to certified 98 percent accurate by NIST (supposedly a high enough rate to tackle bias).
The Virginian-Pilot found that Clearview AI was emailing police officers directly to offer them free trials and the Virginia Mercury quotes the Virginia Public Access Project stating Clearview AI has retained six lobbyists in the state to with the aim of “making it legal for law-enforcement agencies and other public safety entities to use” the technology.
The repeal seems to be making good progress. No one has testified against the legislation in the state House or Senate. Police have already begun installing surveillance cameras in Charlottesville, according to the Mercury, though it does not specify if they are intended for use with facial recognition software.
London Metropolitan Police resume live facial recognition
London Met has carried out its first overt live facial recognition operation for two years, leading to four arrests, reports the force. On 28 January, police set up LFR camera vans around central London’s Oxford Circus primed with a watchlist of 9,756.
Eleven alerts were triggered. One was false, ten were true, of which seven were followed up, leading to the four arrests. Biometric Update contacted the Met on the three alerts not investigated, plus other issues, but has not received a reply.
A larger watchlist than ever before, the false alert rate dropped from 0.08 percent on the last operation in February 2020, to 0.008 percent this time.
Three of the men were wanted on warrants: one for extradition in relation to drug offences and serious assault, fraud and being unlawfully at large; one for a traffic offence; one for drug offences; one for alleged threats to kill.
Silkie Carlo, director of civil liberties group Big Brother Watch was present and told Computer Weekly she saw four stops of which two should not have happened. One was due to outdated data and another to straightforward misidentification. The misidentified was a young Black person who had his fingerprints taken at the operation.
US police training instructor encouraging officers to use facial recognition at traffic stops
An instructor from the Street Cop Training company has advised officers to use facial recognition at traffic stops to run their identity against systems to see if there is a warrant out, even if it is unclear if the person committed a crime while driving, reports Business Insider.
The recommendations were given in a July 2021 podcast aimed at police officers, and included the suggestion of PimEyes. Such approaches may not be legal if officers do not have a reasonable suspicion that the person pulled over has committed a crime, according to a lawyer quoted by Business Insider.
Documents from Street Cop Training obtained by Business Insider also recommend officers use Clearview AI, Thorn Spotlight and China-based FacePlusPlus (also known as Megvii).
Facial recognition part of ‘homeless management system’
The system includes an app for officers, an app for homeless people where they can create a profile, and links to a bed availability system for hostels. Marketing describes how the facial recognition can identify an individual to then give a police officer or first responder details such as likelihood of having needles, warrant status, arrest history or aggression.
For the homeless person app, local businesses can offer vouchers.
Erik J. McCauley, founder and CEO of Odin Intelligence LLC, explained how the system works in an email to Biometric Update: “All Odin products are internal systems designed to be used by local agencies such as Police, Fire, EMS, Social Workers etc. Subjects are entered into the system by the first responder, at the time of the initial call for service or contact.
“The homeless person can opt in by acknowledging our EULA when they download the app for the homeless which provides service referrals, communication, and other notifications and bulletins. Officers use ODIN Face ID only for identification purposes.”
The facial recognition tool is based on Amazon’s Rekognition engine, says the CEO.
McCauley explained how mental and physical disability among the homeless was the reasoning behind incorporating facial recognition into the system: “With up to 70 percent of any homeless population being either physically or mentally disabled, the facial recognition was built at the request of officers who routinely approach folks who are limited functioning, and unwilling or unable to provide identifying information to first responders.”
Biometrics for emergency situations is a growing area for the sector.