San Francisco bans city use of facial recognition

Face recognition reached American police departments as a product before it reached them as a policy. Vendors sold matching services to city and state agencies, and the agencies bought them under ordinary procurement rules, without a public vote on whether the technology should be used at all. The accuracy of those systems was contested: in July 2018 the ACLU of Northern California ran a test in which Amazon’s Rekognition service falsely matched 28 members of Congress to police mugshots, with people of color over-represented among the errors.

On 14 May 2019 the San Francisco Board of Supervisors voted 8-1 to pass the Stop Secret Surveillance ordinance. It does two things. It bars city departments, the police among them, from using facial-recognition technology. And it requires the board to approve the acquisition of other surveillance tools, such as automated license-plate readers and body cameras, before a department may buy them, together with a published policy for their use. The second provision is the broader of the two: it converts a purchasing decision into a public one.

The ban is narrower than the headlines suggested. It applies to city agencies and not to private businesses or individuals, so a shop may still run the technology on its customers. It does not reach San Francisco International Airport or the port, both of which fall under federal jurisdiction. What it governs is the use of the technology by a government on the people it governs.

That was enough to settle the question of whether such an ordinance was possible. San Francisco was the first major American city to bar its own government from using face recognition, and other cities drew on the text when writing their own restrictions in the months that followed. The technology itself kept improving through the same period. The argument it started was no longer about whether the matching worked, but about who may point it at whom.