Shareholders leery of Amazon system

Vote urged on facial-recognition tech

Facial recognition software is coming under increasing scrutiny from civil liberties groups and lawmakers. Now Amazon, one of the most visible purveyors of the technology, is facing pressure from another corner as well: its own shareholders.

As part of Amazon’s annual meeting in Seattle on Wednesday, investors will vote on whether the tech giant’s aggressive push to spread the surveillance software threatens civil rights — and, as a consequence, the company’s reputation and profits.

Shareholders have introduced two proposals on facial recognition for a vote. One asks the company to prohibit sales of its facial recognition system, called Amazon Rekognition, to government agencies, unless its board concludes that the technology does not facilitate human-rights violations. The other asks the company to commission an independent report examining the extent to which Rekognition may threaten civil, human and privacy rights, and the company’s finances.

“This piece of equipment that Amazon has fostered and developed and is really propagating at this point doesn’t seem to us to be in the best interest of the common good,” said Sister Pat Mahoney, a member of the Sisters of St. Joseph, a religious community in Brentwood, N.Y., that is an Amazon investor and introduced the proposed sales ban. “Facial recognition all over the place just makes everyone live in a police state.”

The proposals are nonbinding, meaning they do not require the company to take action, even if they receive a majority vote. But they add to the growing resistance to facial surveillance technology by elected officials, civil liberties groups and even some Amazon employees.

Last week, San Francisco banned the use of facial surveillance technology by police and other city agencies. Oakland, Calif., and Somerville, Mass., near Boston, are considering similar bans. Earlier this year, state lawmakers in Massachusetts and California introduced bills that would restrict its use by government agencies. On Wednesday, the U.S. House Committee on Oversight and Reform will hold a hearing on the civilrights implications of facial surveillance.

For Amazon’s annual meeting Wednesday, employees who are stockholders have also introduced a proposal on climate change, pushing the company to make firm commitments to reduce its carbon footprint.

But Amazon fought particularly hard to prevent the votes on facial surveillance. In a letter to the Securities and Exchange Commission in January, the company said that it was not aware of any reported misuse of Rekognition by law enforcement customers. It also argued that the technology did not present a financial risk because it was just one of the more than 165 services Amazon offered.

“The proposals raise only conjecture and speculation about possible risks that might arise” from clients misusing the technology, lawyers for Amazon wrote in the letter. The agency disagreed, ultimately requiring Amazon to allow the facial surveillance resolutions to proceed.

In a statement in response to a reporter’s questions, Amazon said it offered clear guidelines on using Rekognition for public safety — including a recommendation that law enforcement agencies have humans review any possible facial matches suggested by its system. The company added that its customers had used Rekognition for beneficial purposes, including identifying more than 3,000 victims of human trafficking.

“We have not seen law enforcement agencies use Amazon Rekognition to infringe on citizens’ civil liberties,” the Amazon statement said.

The New York Times used Amazon Rekognition last year to help identify guests at the royal wedding of Prince Harry and Meghan Markle.

Amazon is becoming a national magnet for mounting opposition to facial surveillance — a technology that may be used to identify and track people at a distance without their knowledge or consent.

Facial recognition uses artificial intelligence to scan a photo of an unknown person. The software then compares the facial template of the unknown person with a database of templates of known people and, if the templates are very similar, may suggest a name or match.

Proponents of the technology argue that such systems help law enforcement agencies more easily identify crime suspects and missing children. Civil liberties groups warn that the technology could easily be misused to disproportionately pursue immigrants, people of color and protesters, infringing on their rights to free speech and movement.

Other companies have long sold facial surveillance to law enforcement agencies, but Amazon has differentiated itself by, in part, playing down warnings about the technology.

Last year, Google said that it would refrain from offering facial recognition for general purposes until it had worked through the policy implications. This year, Bradford Smith, the president of Microsoft, said that his company had decided not to sell the surveillance technology to a police department seeking to freely use it on the general public.

Amazon, in contrast, recently pitched its facial recognition services to U.S. Immigration and Customs Enforcement, according to company emails obtained under open records law by the Project on Government Oversight, a nonprofit group based in Washington.