Amazon facial recognition technology misidentifies 28 lawmakers as those previously arrested, ACLU says

Law enforcement agencies around the country are considering using the software.

July 26, 2018, 9:54 PM

The American Civil Liberties Union (ACLU) claims that in a test it conducted, Amazon’s facial recognition technology falsely matched 28 members of Congress with other people who had been arrested for a crime.

In a report published Thursday, the ACLU said it input the photos of every current member of the House and Senate into a database it had created of 25,000 publicly available mugshots. Using the default settings on Amazon’s Rekognition facial recognition system, the ACLU says more than two dozen congresspersons were misidentified as someone who had been arrested.

“We remain excited about how image and video analysis can be a driver for good in the world, including in the public sector and law enforcement,” a spokesperson with Amazon Web Services said in a statement.

The statement also said that the technology is “almost exclusively used to help narrow the field and allow humans to expeditiously review and consider options using their judgment.”

PHOTO: Amazon corporate office building is seen in Sunnyvale, Calif.
Amazon corporate office building is seen in Sunnyvale, Calif.
Moment Editorial/Getty Images

Police departments around the country have begun to consider using facial recognition technology. The technology recently helped authorities in Maryland identify Jarrod Ramos, the alleged gunman in the Capital Gazette shooting.

ACLU’s test included both Democratic and Republican senators and house representatives of all ages and genders.

Their report suggested a racial bias in the technology’s errors. Of the members of Congress misrepresented, 39 percent of false matches were people of color. However, people of color only make up 20 percent of Congress.

Rep. Luis Gutierrez, D-Ill., and Rep. Jimmy Gomez, D-Calif., both of whom were misidentified in the test, have responded on Twitter.

Gomez tagged Amazon CEO Jeff Bezos asking to talk and Gutierrez tweeted, “Are you sure Amazon isn’t just talking to Donald Trump, Tucker Carlson, and Breitbart? They think all Latinos are criminals.”

Gutierrez, along with Sen. Edward J. Markey, D-Mass., and Rep. Mark DeSaulnier, D-Calif., both of whom had also been misidentified by the technology sent a letter to Bezos with a series of questions about how the technology works.

In a press release, civil rights leader Rep. John Lewis, D-Ga., who had also been misidentified, called the test’s results “deeply troubling.”

PHOTO: Rep. John Lewis speaks at the dedication of the Smithsonian's National Museum of African American History and Culture in Washington, D.C., Sept. 24, 2016.
Rep. John Lewis speaks at the dedication of the Smithsonian's National Museum of African American History and Culture in Washington, D.C., Sept. 24, 2016.
Joshua Roberts/Reuters FILE

“I myself have suffered misidentification and erroneous targeting while traveling through airports and on airplanes as I commute between D.C. and Georgia,” Lewis said. “What would happen, under these already threatening conditions, if people from these same communities were misidentified by facial recognition software? How would they prove to [the] police that a computerized result is false?”

Lewis also said in the release that law enforcement agencies should not use the technology until “the onerous civil rights and civil liberties issues are confronted and accuracy is guaranteed.”

The Congressional Black Caucus sent a letter to Bezos in May expressing their concern about the company working with law enforcement agencies to implement facial recognition technology.

At the end of the report, the ACLU called for Congress to “enact a moratorium” on law enforcement agencies using facial recognition technology.

Related Topics