U.S. lawmakers aim to curtail face recognition even as the technology IDs Capitol attackers

[January 19, 2021]  By Paresh Dave

OAKLAND, Calif. (Reuters) - U.S. lawmakers are moving ahead with efforts to ban facial recognition software even as the technology helps identify supporters of President Donald Trump who ransacked their workplace and forced them to evacuate this month.

Researchers and law enforcement have been running photographs from the Jan. 6 siege of the Capitol through facial recognition, which looks for similar faces in databases of mugshots, social media headshots or other images.

"It's a great tool," said Michael Sheldon, research associate at the nonprofit Atlantic Council's Digital Forensic Research Lab, whose mission includes protecting democratic institutions.

Sheldon said he provided the FBI possible names of several people in the violent mob thanks to a facial recognition program he purchased.

Senator Ed Markey said in a statement that the technology may not be worth the risks. Racial justice activists have warned that facial recognition can perpetuate discriminatory policing and constant tracking could become the norm.

"We've seen that in the wake of past crises innocent Americans have been subjected to increased problematic surveillance," Markey said.

He called on authorities to "keep the public safe and hold criminals accountable without relying on invasive tools that are proven to have serious accuracy and bias issues."

Markey last year joined three other Democrats in Congress to introduce an unsuccessful bill that would have banned federal agencies such as the FBI from using facial recognition.

Chris Evans, spokesman for Pramila Jayapal, one of the other lawmakers behind that effort, said members plan to reintroduce the proposal this year.

Whether the FBI has used facial recognition in the ongoing probe is unclear. It declined to comment on its tools.

U.S. National Guardsmen stand duty behind the unscalable fence newly topped with razor wire surrounding the U.S. Capitol in the wake of the January 6th riot and ahead of the upcoming inauguration in Washington, U.S. January 14, 2021. REUTERS/Jonathan Ernst/File Photo

Most of the nearly 100 assailants criminally charged so far turned themselves in, were outed by acquaintances or posted about their allegedly illicit activities on social media, according to court records.

But people have sent the FBI tips based on facial recognition. Miami police submitted 13 possible names from searches it ran using software from Clearview AI, Assistant Chief Armando Aguilar told Reuters.

Sheldon said the tool he used, which he declined to publicly name, looks for online matches, such as on websites that aggregate mugshots or university alumni lists. He doublechecked likely hits using facial recognition from Microsoft Corp's Azure cloud computing service, he said.

He said the software linked a man photographed carrying zip-tie handcuffs in the Capitol to a headshot on Alumnius.net under the name Larry Brock.

Prosecutors charged Brock with violent entry and other counts following a New Yorker article, in which researchers identified him and he acknowledged being at Capitol. Brock's attorney did not respond to a request for comment.

Facial recognition previously helped Sheldon unmask mercenaries involved in an attack in Syria as part of research into global conflicts and disinformation.

For the Capitol investigation, the technology enabled him to quickly find assailants on social media and study their motivation.

(Reporting by Paresh Dave; Editing by Jonathan Weber and Bill Berkrot)

[© 2021 Thomson Reuters. All rights reserved.]

Copyright 2021 Reuters. All rights reserved. This material may not be published, broadcast, rewritten or redistributed.  Thompson Reuters is solely responsible for this content.