New facial recognition to add more layers to Mall of America security
Like this?
Log in to share your opinion with MPR News and add it to your profile.
Like this?
Thanks for liking this story! We have added it to a list of your favorite stories.
The Mall of America implemented new facial recognition technology into their security system on Wednesday.
This advanced system will “provide another layer of security” for the 5.6 million square foot building, the mall said in a statement.
“Safety is our number one priority for our guests, tenants and employees. We are constantly looking at ways to evolve our security protocols and our programs and practices, and this is just another additional layer to our already comprehensive security program,” Will Bernhjelm said, the vice president of security.
The mall said the facial recognition will only be used for the purpose of detecting persons of interest. That includes people who are considered to be trespassing at the Mall of America, people who may be a threat, people identified by law enforcement, or people who are missing or may be in danger.
Support the News you Need
Gifts from individuals keep MPR News accessible to all - free of paywalls and barriers.
The technology will upload photos of the persons of interest into the facial recognition system, which looks for matches based on the mall’s database. If there is a match, the software system will alert security, who performs traditional facial recognition to see if the person is a match.
According to Bernhjelm, no data is stored on individuals who are not matches.
“If you’re on a trespass, you shouldn’t be here, and you shouldn’t be allowed access to the building. And we’re hoping that this helps keep those people out,” Bernhjelm said. “We don’t care who the regular shoppers are. We’re not out scanning their face and then sending that picture out trying to compare it to images that are on the internet and figure out who they are and their name and where they’re from.”
Once an alert is issued, security conducts an investigation that includes three layers of “traditional” facial recognition.
The mall’s security team was trained by Ideal Innovations on traditional facial recognition, where they learned from experts about facial morphological features and how to successfully compare faces to determine whether the system alert is accurate.
“We’ve done a lot to enhance our security program and it’s always constantly evolving. I have worked here for 21 years. It is like home, right? And we kind of protect it like that. Not only is this a job, but it’s also a place where I bring my family. So it is very important to me that we're doing the right thing and protecting it in a way that is leading edge,” Bernhjelm said.
If someone has been found to be a match after their traditional facial recognition, a Mall of America officer will approach and follow the standard security procedure when addressing the potential person of interest. This is only after human review has determined with sufficient confidence that the alert is accurate.
The mall has been talking about these changes for nearly two years. They decided to use Corsight, an Israel-based company, as their facial recognition company. Corsight’s technology is powered by Autonomous AI.
Minneapolis voted in 2021 to ban the use of facial recognition software, saying it could be used to increase surveillance in communities of color.
“This would further disproportionately harm communities that historically have faced elevated levels of policing and harassment,” the ordinance said.
The city was also skeptical of the technology’s accuracy.
“Facial recognition technology has been shown to be less accurate in identifying people of color and women. Facial recognition technology has the potential to further harm already disadvantaged communities through incorrect identifications,” the ordinance said.
According to the mall, the algorithm correctly identified individuals 99.3 percent of the time in Department of Homeland Security (DHS) testing. The facial recognition technology is powered by an algorithm that has undergone testing from the National Institute of Standards and Technology (NIST) and the DHS, Bernhjelm said.
“The algorithm performed exceptionally well,” Bernhjelm said. “I’m not shy to say that we have a best-in-class system.”
The price of the new technology has not been revealed, but Bernhjelm said they’ve “invested a lot of time and resources and money into supporting our security goals.”