The technology will only be used to track persons of interest, which includes people "who are currently on a trespass at Mall of America, those who may be a threat to our environment, persons identified to us by law enforcement, or individuals who are missing or may be in danger," the mall said in a statement.
Photos of individuals of interest are uploaded to the recognition software and the technology looks for matches to those photos. If there is a match, the system alerts the security team, which launches a further investigation, the Bloomington-based mall said.
The tech does not track or keep information on people who are not of interest, the mall said.
"At 5.6 million square feet, our officers cannot be everywhere at once," said Will Bernhjelm, vice president of security at Mall of America. "Utilizing this cutting-edge technology will allow us to more quickly do what we are already doing: identifying individuals of interest and keeping Mall of America and its guests safe."
Mall security is trained in traditional facial recognition to learn how to compare facial features to determine whether the software alert is a match. Following human review, mall officers will approach a potential person of interest and follow standard security procedures.
Facial recognition technology has been shown to misidentify people of color at higher rates. AI-based facial recognition has lead to wrongful arrests and have poorest accuracy rates with persons who are young, Black and female.
In nearby Minneapolis, the technology was banned for use by police and other city agencies in 2021.
The mall's vendor, Corsight AI, has been tested by the National Institute of Standards and Technology (NIST) and the Department of Homeland Security (DHS).
"It is by all accounts, best in class. In the DHS testing, the algorithm correctly identified individuals 99.3% of the time," the mall said.
©2024 StarTribune. Visit startribune.com. Distributed by Tribune Content Agency, LLC.
