Skip to main content

Amazon told employees it would continue to sell facial recognition software to law enforcement

Amazon told employees it would continue to sell facial recognition software to law enforcement

Share this story

Illustration by James Bareham / The Verge

Amazon has come under fire of late for the licensing of its controversial Rekognition system, a powerful piece of facial recognition software, to government and law enforcement agencies, with the most recent development involving revelations that Amazon met with US Immigration and Customs Enforcement (ICE) over the summer to strike a deal for Rekognition use. Now, Amazon employees are grilling the company’s leadership over the selling of such technology, especially when it could be used to track human beings and send them back into potentially dangerous environments overseas.

Amazon Web Services CEO Andrew Jassy told employees at an all-hands meeting today that, “We feel really great and really strongly about the value that Amazon Rekognition is providing our customers of all sizes and all types of industries in law enforcement and out of law enforcement,” according to an Amazon employee who spoke with The Verge under the condition of anonymity and provided a partial transcript of the conversation. Gizmodo first reported on Jassy’s comments earlier this afternoon.

Amazon was not immediately available for comment.

AWS, the cloud computing division of Amazon, is responsible for powering Rekognition, which is currently in use by police departments in Florida and Oregon. Jassy was tapped by Amazon CEO Jeff Bezos to specifically answer a question about Rekognition and potential government contracts with organizations like ICE. Jassy went on to say that “you don’t want to get rid of that technology,” but noted how violations of AWS’ terms of service or violations in general of constitutional rights would force Amazon to ban use of its services. He did note, however, that the company believes “it’s the role in the responsibility of the government to help specify what the guidelines of regulations should be about technology.”

Compounding Amazon’s participating in government contracts with Rekognition is research proving the system is deeply flawed, both in terms of accuracy and regarding inherent racial bias. The American Civil Liberties Union tested Rekognition over the summer and found that the system falsely identified 28 members of Congress from a database of 25,000 mugshots. (Amazon pushed back against the ACLU’s findings in its study, with Matt Wood, its general manager of deep learning and AI, saying in a blog post back in July that the data from its test with the Rekognition API was generated with an 80 percent confidence rate, far below the 99 percent confidence rate it recommends for law enforcement matches.)

The company has also declined to participate in a comprehensive study of algorithmic bias run by the National Institute of Standards and Technology that seeks to identify when racial and gender bias may be influencing a facial recognition algorithm’s error rate.

Amazon isn’t the only technology giant experiencing pushback from its own employees about how products are sold to and used by the US government. Google leadership has faced immense backlash this year over its participation in a Pentagon drone initiative known as Project Maven, forcing the company to announce it would not renew its contract next year and to publicize a set of ethics guidelines around artificial intelligence use and product development. Google is also coming under fire for its plans to launch a news and search product for the Chinese market that would conform to the government’s censorship policies.

Here’s the full transcript of the exchange from the all-hands meeting:

Q: What is being done in response to the concerns voiced by both Amazon employees and civil rights groups regarding Amazon selling facial recognition technology to government and police organizations, including ICE?

Amazon CEO Jeff Bezos: Good. Andy, do you want to take that one?

AWS CEO Andrew Jassy: Thank you, so we’re referring to Amazon Rekognition, which is AWS’ deep learning image recognition, facial recognition, video recognition service and, you know, with over five hundred thousand employees like we have at Amazon, I think we’re going to have people who have opinions that are very wide-ranging, which is great. But we feel really great and really strongly about the value that Amazon Rekognition is providing our customers of all sizes and all types of industries in law enforcement and out of law enforcement.

You see it in the value people are actually getting even after just about a year, year and a half of the service, where Rekognition is actively been used to help stop human trafficking, to reunite missing kids with parents for educational applications, for security and multi-factor authentication to prevent theft.

There’s a lot of value being enjoyed from Amazon Rekognition. Now now, of course, with any kind of technology, you have to make sure that it’s being used responsibly, and that’s true with new and existing technology. Just think about all the evil that could be done with computers or servers and has been done, and you think about what a different place our world would be if we didn’t know how people have computers.

So you don’t want to get rid of that technology. You want to make sure that people use the technology responsibly, and we have a set of terms and services in AWS. And with all our services, including Rekognition, where if people violate those terms of services and don’t use them responsibly, they won’t be able to use our services any longer. In fact, if we find the people are violating folks constitutional rights, they won’t be able to use the services. I also think, by the way, in a democracy is also often the role in the responsibility of the government to help specify what the guidelines of regulations should be about technology. And, if and when that happens, we will abide by those as well. Thank you.

Update 11/8, 7:16PM ET: Added information about research into the flaws of Amazon’s Rekognition system.

Correction: An earlier version of this article incorrectly stated that BuzzFeed had first reported details around Amazon’s all-hands meeting. That is incorrect; Gizmodo first published the news.