US Senators seek balance between facial recognition benefits, risks for investigations

The U.S. Senate Judiciary Committee dedicated a large portion of its latest hearing on the use of artificial intelligence in criminal investigations to exploring the risk of discrimination by law enforcement using facial recognition.

Senator Cory Booker (D-NJ) noted the “astonishingly low” rate of solved murder investigations in the United States, around 50 percent. Senator Tom Cotton (R-AR) referred to a statement from the Major County Sheriffs of America and the Major Cities Chief Association in arguing that facial recognition can serve as an invaluable “investigative assistant to law enforcement rather than a replacement for the human element.” Cotton provided an example of a child abuser caught with facial recognition in mid-2023.

Miami Police Department Assistant Chief Armando Aguilar testified that a study from Florida International University shows a much higher rate of solved cases when detectives use AI tools. He also spoke at length about how his police force’s facial recognition policy was formulated, and the emphasis it places on transparency in order to address some of the common concerns around the technology.

Government Accountability Office Director of Science, Technology, Assessment, and Analytics Karen L. Howard reviewed GAO findings that AI applications in latent fingerprint biometric analysis and facial recognition offer significant advantages over human analysts. The GAP has recommended increased training for analysts and investigators, support for appropriate use standards, and increased transparency around the testing and use of algorithms.

Senators turned to questions around what steps have been taken to mitigate the risk of bias in facial recognition systems, and the role of NIST testing in avoiding that risk.

The social and political context for the hearing was set by ongoing revelations about police misconduct in cases where facial recognition is used and people are wrongly arrested, most of them Black.

Despite this, Rank One Computing CRO and SIA Identity and Biometric Technology Board Chair Benji Hutchinson says the hearing was more balanced than many in the past.

“There seems to be more curiosity both about the challenges facing the use of AI technology but also the positive outcomes,” Hutchinson told Biometric Update in a message over LinkedIn.

“The technologies themselves are not the issue,” Berkeley Center for Law and Technology Co-director Rebecca Wexler said as the hearing concluded. “It’s the legal rules that we set up around them to help us ensure that they are the best, most accurate and effective tools and not flawed or fraudulent in some way.”

Members of Congress learning

“Overall, this hearing was a step in the right direction for the biometrics and identity industry,” says Hutchinson, who is also an appointee of the National Artificial Intelligence Advisory Committee’s Subcommittee on Law Enforcement. “This event coincides with a number of other studies and hearings around AI tools in law enforcement, that include facial recognition. We see numerous committees in the House and Senate eager to learn more about facial recognition and its applicability to law enforcement. The National Academy of Sciences recently released their study on the technology. Members of Congress and their staffers are getting much smarter on these topics and so is the general public. NIST is overseeing the NAIAC Law Enforcement Sub-Committee that is developing recommendations for the full NAIAC and White House on these issues.”

“Another critical take away is that citizens and neighborhoods affected by violent crime overwhelming welcome these AI tools in their community,” he observes. “This is a narrative we don’t often hear about.”

“We are learning that when used responsibly, as in Miami, these tools are quite effective in combating violent crime,” Hutchinson adds. “Miami was one of the most violent cities in America in the 1980’s. We learned that the murder rate has plummeted and clearance rates for violent crimes have increased dramatically, due in part to the effective implementation of these AI tools. We still have work to do specifically around providing training to officers and ensuring these tools are used as intended to produce leads.”

Article Topics

biometrics  |  criminal ID  |  facial recognition  |  police  |  ROC  |  United States