By The Washington Post 
30m ago
By Drew Harwell
Amazon is extending its global ban on police use of its facial recognition software until further notice, the company said Tuesday, prolonging a one-year moratorium on a surveillance technology that has stirred controversy due to its problems with racial bias and false arrests.
The tech giant said in June, amid nationwide protests over racial injustice and police violence, that it was instituting a one-year police ban of its software, Rekognition, to “give Congress enough time to implement appropriate rules” governing the technology’s ethical use.
On Tuesday, the company said it was extending the ban indefinitely. An Amazon spokesperson declined to provide further detail. (Amazon CEO Jeff Bezos owns The Washington Post.)
Although lawmakers have proposed some limited federal regulations in the months since, Congress has yet to pass laws related to police use of such technology. More than a dozen cities and states have enacted their own laws banning or restricting the technology’s use by public officials or local police.
Council members in King County, Wash., where Amazon’s Seattle headquarters is based, are considering a ban this month. And in Virginia, where Amazon is building its second headquarters, known as HQ2, state lawmakers enacted one of the strictest facial recognition laws in the country, requiring local law enforcement to secure state legislative approval before using any facial recognition system.
Rekognition had been sold to local police departments as a way for officers to upload an image of a potential suspect and quickly scan for matches across a database of hundreds of thousands of jail mug shots or other photos.
Its first client, the Washington County Sheriff’s Office in Oregon, told The Washington Post in 2019 that it had run more than 1,000 searches through the system in the previous year, including photos from store surveillance cameras and Facebook profiles.
The technology has been criticized for its potential for error, particularly after researchers found that Amazon’s system, like other facial recognition software, was more likely to return false matches when analyzing people of color.
Police have been found to use edited photos, sketches and other distorted images while searching for potential suspects, raising the risks of faulty results.
Three men, all of whom are Black, have sued U.S. police departments for false arrests tied to facial recognition mismatches. All three cases are ongoing.
Amazon is one of many companies that has marketed facial recognition technology for sale: Hundreds of similar facial recognition algorithms have been developed, some of which are being promoted for police, commercial or personal use. IBM and Microsoft stopped selling the technology to police last year.
The FBI and other federal agencies have run hundreds of thousands of facial recognition searches in the last decade, government data show. Some companies now allow anyone to run facial recognition searches.
The Washington Post