Amazon offered ML facial recognition service to ICE

Recently unveiled emails show Amazon pitched the use of Rekognition, a facial recognition algorithm, to assist ICE agents identify illegal immigrants

24Oct

A new report has revealed that Amazon met with the US Immigration and Customs Enforcement (ICE) agency in June 2018 to pitch the use of its facial recognition machine-learning platform, Rekognition.

The meeting was brought to light after emails were uncovered by Andrea Peterson of the Project on Government Oversight, a body that investigates the federal government's dealings in an effort to expose fraud, waste and other conflicts of interest.

According to Bloomberg, Amazon representatives showcased a number of machine-learning tools contained within the Amazon Web Services platform. However, the focus of the meeting was to pitch Rekognition to ICE officials, with the platform offering quick identification of illegal immigrants from images or videos.


Visit Innovation Enterprise's Machine Learning Innovation Summit, part of DATAx New York on December 12–13, 2018


Mathew Bourke, a spokesperson for ICE, commented that the agency regularly meets up with new vendors to explore new tools which have been developed. Since the meeting, no contract has since been reached, although Amazon does have contracts with smaller governmental bodies that have a monthly subscription to Amazon's Rekognition tool.

The use of facial recognition software by the US government has been frowned upon by critics who say it has the potential to infringe on personal civil liberties. Also, with regular accounts of biased AI being reported, another pertinent criticism is that the US government should not be using an algorithm which has not been submitted to the National Institute of Standards for authenticity and racial bias testing.

Commenting on the report, an Amazon spokesperson said: "We participated with a number of other technology companies in technology boot camps sponsored by McKinsey Company, where a number of technologies were discussed including Rekognition."

How to take the risk out of saas development small

Read next:

How to take the risk out of SaaS development

i