Microsoft Board Opposes Banning Facial Recognition Sales: Proxy

A Microsoft shareholder proposal up for a vote at the Nov. 30 virtual annual meeting asks the board to “generally prohibit sales” of facial recognition technology to law enforcement and all other “government entities,” and to publicly disclose “any exceptions made,” said a proxy statement Thursday filed at the SEC. The board recommends voting against the proposal, saying the requested policy “does not advance the interests of Microsoft” or its shareholders and other stakeholders.

Shareholders worry Microsoft’s facial recognition technology “poses risk to civil and human rights and shareholder value,” says the proposal. It cites a June report in which 50 global investors representing more than $4.5 trillion in assets urged companies involved in the development and use of facial recognition technology “to proactively assess, disclose, mitigate and remediate human rights risks related to the technology.”

Microsoft committed in 2020 that it won’t sell facial recognition technology to U.S. police departments before strong regulation, “grounded in human rights,” has been enacted, says the proposal. But Microsoft’s policy “does not address potential sales to local, state or federal agencies, or to governments outside the U.S., with a history of oppression and human rights abuses,” it says.

The company needs to align with its corporate peers on facial recognition sales, says the proposal. “A shareholder proposal at Amazon addressing concerns associated with the use of facial recognition software by government and law enforcement has garnered increasing shareholder support over the past three years,” it says. IBM announced in 2020 that it would no longer offer general purpose facial recognition technology over concerns about mass surveillance, racial profiling and human rights violations, it says.

At least 23 U.S. municipalities since 2018 have adopted legal bans on facial recognition technology, including King County, Washington, where Microsoft is headquartered, says the proposal. Local bans, “building on the evidence of concern, have demonstrated that facial recognition technologies in the hands of government are an inherent threat to civil and human rights,” it says. Microsoft “should acknowledge this by banning sales to government entities,” it says.

The board disagrees, says the proxy, citing “extensive public commitments” Microsoft has made “to restrict our sale of facial recognition technology based on human rights considerations.” The shareholder proposal “does not recognize those commitments and would impose a blunt prohibition that would deny public agencies the ability to deploy facial recognition technology in societally beneficial use cases,” it says.

There's no one-size-fits-all approach to facial recognition technology, says the board. “Different use cases have different risk profiles, and safeguards need to be calibrated accordingly,” but the shareholder proposal “draws no such distinction,” it says. The proposal, for example, “would restrict the ability of Microsoft to provide government customers with the Windows Hello feature, which utilizes facial recognition technology to help users to more securely sign onto their devices, at a time when governments face punishing ransomware and other cyberattacks,” it says.

Microsoft’s “responsible AI program” has a “demonstrated track record of articulating and operationalizing our AI principles in ways that address the human rights concerns cited in this proposal,” says the board. “Microsoft continues to advocate for new regulations for responsible facial recognition use that are grounded in human rights protections,” by addressing testing, transparency and “accountability requirements,” it says.