Axon says it won't make police body cams with facial recognition tech

Axon, the company formerly known as Taser International, said in a statement this week that it will not produce police body cameras that feature facial recognition technology. The company, which refers to this technology as 'face matching,' made its decision based on a recommendation from its Axon AI and Policing Technology Ethics Board.

Axon formed its ethics board in April 2018, and has since published its first board report with recommendations covering facial recognition technology. Axon says that it agrees with these recommendations and will not be implementing face matching tech in police body cameras 'at this time.'

Facial recognition technology has proven highly controversial with a number of advocacy groups and experts raising concerns over potential abuses of the technology, as well as ways in which the technology fails to accurately recognize faces and other related issues.

Despite these warnings, the technology has made its way into a huge number of public places, including airports where it is increasingly used to verify passenger identities, as well as private and law enforcement surveillance systems. The technology has proven useful for automating certain tasks, such as automatically finding and blurring faces in footage.

Use of the technology as part of a system that matches faces to an identity database has raised a number of concerns, and Axon says that in light of those various issues, it will not be adding this type of system to its body cameras. Beyond the ethical concerns, Axon points out that there are also multiple 'technological limitations' associated with face matching on body cameras.

The company's language indicates that future body cameras featuring face matching technology haven't been ruled out, but Axon didn't shed light on whether it anticipates eventually bringing this type of body camera to market in the future.

At this point in time, Axon's facial recognition work has been limited to systems that find faces in videos for the purposes of blurring them out or redacting them. This is distinctly different from systems that use face matching with a database for identification purposes.