ISC West Day 2 Keynote Recap: Will Bernhjelm

The story of retailing in America is, in many ways, the story of America. It’s no surprise, then, that the largest physical retail space in the country is named The Mall of America. Located in Bloomington, Minn., The Mall of America comprises 5.6 million square feet of retail, lodging and transportation facilities. Its 32 million visitors each year come and go through 27 entrances and exits. Keeping store customers, travelers and employees safe is an enormous undertaking.

Will Bernhjelm, the person responsible for securing that space, spoke to a capacity audience on the main stage at ISC West recently about the challenges he faces and some of the tactics he employs to meet those challenges. Layering different resources and technologies is the only way to protect a facility that large with that much access available to that many people.

“We have a very robust program, with various layers built in. Some of those layers are visible, others are not,” he explained. “I use the analogy of Swiss cheese. Within every layer of security, like Swiss cheese, there’s a hole, a vulnerability. But the more layers we have stacked on one another, the less chance that all those holes show up at once.”

For Bernhjelm and the Mall of America, the layers he uses include a human intelligence analyst looking for threats—mostly online—before they show up on the property, an extremely close relationship with the Bloomington Police Department, exterior bike patrols, an extensive array of cameras, a staff of highly trained uniformed patrol, an in-house canine unit, a ban on unaccompanied minors and, as of June 2024, facial recognition technology.

Berhjelm spent significant time during his keynote address explaining the company’s philosophy, compliance efforts and narrow scope of use. Bernhjelm is clear that the system only compares people walking in the door to a database of images of “persons of interest”—mostly people who have trespass notices already in force. It does not, he says, store or save images that the cameras capture within the Mall of America.

“Also, any alert the system throws triggers a three-level human review,” he says. “As good as it is (and it has over 99 percent accuracy), we are not going to trust the tech alone. The dispatch team that gets the alarm does their own comparison. It then goes to the officers on the floor on their mobile phones, who will also compare the image to the person. We have hammered into the staff over and over, that just because we get a notification, they need to verify that it’s the right person.”

It is the narrow purpose of the technology that enables Bernhjelm to effectively use it while complying with a fluid regulatory situation. If a subject is not a previously identified person of interest, the Mall of America does not retail any images or information in the facial recognition system. To close out his talk, Bernhjelm presented data on the number of trespass stops since implementing the system. The number of stops Mall officers were able to make with the new system nearly doubled from the old method of humans needing to remember and recognize the trespassers. Part of the decrease, he says, is a deterrence effect. Potential trespassers are communicating on social media.

“The word,” he says, “is getting out.”