In the ongoing fight for a better, fairer criminal justice system, truly admirable are the victims of terrible, government-promoted unfairnesses who refuse to back mere half-measures aimed at a fix.
Such is the case with Robert Williams of Detroit, who has come to California to lobby against a well-intentioned bill in the California Legislature crafted to help contain police use of facial-recognition technology, which often fingers the wrong people as suspects — especially when they are Black people.
Williams says the legislation would continue to allow bad arrests, and he wants the proposed law improved, and we agree.
We’ve written about Williams’ plight before. As Khari Johnson of CalMatters recounts, Detroit police in 2020 falsely accused Williams of stealing luxury watches worth thousands of dollars and arrested him in front of his wife and daughters after facial-recognition technology “matched a surveillance video to a photo of Williams in a state database.”
AB 1814 by Assemblyman Phil Ting, D-San Francisco, would indeed make it against the law for California police to use facial-recognition AI as the only reason for a search or arrest, requiring “corroborating indicators” as well.
That’s certainly a step in the right direction. But Williams came to Sacramento to testify that the proposed law would not have stopped his arrest. “In my case, as in others, the police did exactly what AB 1814 would require them to do, but it didn’t help,” said Williams. “Once the facial recognition software told them I was the suspect, it poisoned the investigation.”
We applaud Ting for his extremely well-intentioned efforts at curbing the power of the surveillance state. We agree that setting up guardrails now is important, before police turn everything over to the robots. Research shows facial recognition is more error-prone when it comes not only to Black people, but to Asians and Native Americans as well.
A major problem is that, unlike the tech in your smartphone that recognizes your face, the databases police use contain millions of photos of people from drivers licenses, and, as CalMatters notes, “can fail in numerous ways.” People aren’t particularly good at picking out suspects from photos; neither are machines.