SACRAMENTO, Calif. (CN) – When the American Civil Liberties Union used facial recognition software to cross-check 120 California legislators against a database of 25,000 publicly available mug shots, the algorithm falsely identified 26 of the lawmakers as someone who’s been arrested.
One such lawmaker was Assemblyman Phil Ting, D-San Francisco, was matched with a mug shot by Amazon’s facial recognition software called Rekognition despite never having been arrested.
“This is a demonstration that this software is not ready for primetime,” Ting said during a press conference in Sacramento on Tuesday. Ting joined representatives from civil liberties advocacy groups to announce the introduction of Assembly Bill 1215, which seeks to ban the use of facial recognition software in police body cameras.
Assemblyman Reggie Jones-Sawyer, D-Los Angeles, who was also linked to a mug shot in the database, said facial recognition software in its current iteration is particularly inaccurate when it comes to people of color and women. Researchers at MIT have said Amazon Rekognition is less accurate in identifying anyone who is not a white male.
“Putting this software in body cameras would automate mistaken identity,” Jones-Sawyer said.
Amazon refuted Tuesday’s findings, accusing the ACLU of manipulating the software to cast it in a negative light.
“The ACLU is once again knowingly misusing and misrepresenting Amazon Rekognition to make headlines,” the technology giant said in a statement. “When used with the recommended 99% confidence threshold and as one part of a human-driven decision, facial recognition technology can be used for a long list of beneficial purposes, from assisting in the identification of criminals to helping find missing children to inhibiting human trafficking.”
Matt Cagle, a lawyer with the ACLU Northern California, said the fact that the software produced inaccuracies is troubling but not the only reason the state should ban its use in body cameras.
“Even if they were accurate, we should agree body cameras are not to be used for surveillance,” he said, pointing out the advent of body cameras stemmed from efforts to introduce transparency and accountability between police and communities.
“It’s not a problem that can be fixed by tweaking an algorithm,” he said. “People should be able to attend a protest without having their identity and location recorded and stored in a government database.”
Cities like Oakland and San Francisco have implemented total bans on facial recognition technology.
Ting said the state could contemplate stronger bans but he wanted to start with body cameras because the idea has a greater political feasibility.
“I’ve had good conversations with the governor,” Ting said. Should the bill pass both wings of the statehouse, Governor Gavin Newsom would also have to sign it into law.
New Hampshire and Oregon have passed similar laws outlawing the use of the technology in body cameras.