AI Weapons Detection Startups Compete With Industry Giant in Expanding Virginia School Market

Demand from Virginia school divisions for better building security is on the rise following increases in school shootings over the past several years, including a high-profile case at a Newport News elementary in January.

According to the K-12 School Shooting Database, Virginia had 13 school shootings in 2022, the most on record for the state. As of early August, the number of wounded and killed in 2023 had already matched last year’s total.

This May, Republican Gov. Glenn Youngkin announced $16.4 million in competitive grant funding for additional security infrastructure at schools, and so far at least 10 school divisions in Virginia have plans to install new weapons detection systems. 

(K-12 School Shooting Database)

As school officials and communities deliberate on how to best protect their students, two different approaches to school security are competing for control of the nascent market: more traditional metal detection on the one hand versus artificial intelligence-powered weapons detection on the other.

Over the past decade, advancements in machine learning technology have made it possible for startups like Massachusetts-based Evolv Technology to take on industry staples like Italian company Costruzioni Elettroniche Industriali Automatismi, better known as CEIA, in the market for school contracts in Virginia and across the United States.

Evolv currently has contracts with the Prince William County and Alexandria City school divisions, and spokesperson Jill Lemond said the company is in talks with several others but declined to name them.

“We have schools that range from urban Northern Virginia districts, and then we also have some that are in more rural districts. So it’s not even just one type of school,” Lemond said. “We’re trying to help the whole state.”

Evolv’s key product, the Express screening device, differs from traditional metal detection systems in that it’s able to screen multiple people at the same time and can identify the shapes of objects in addition to what they’re made of. By using computer vision, an AI methodology for identifying objects from images, the technology can pick out a weapon from a crowd passing through the device’s columns.

Virginia schools increasingly eye weapons detectors to keep firearms out of buildings

“What we’re able to do differently is we have an image-aided alert that allows for targeted searches,” said Lemond. “It is looking for not only the metallic value but also the shape of those items that look like parts or pieces of large weapons, mass casualty-type weapons.”

Evolv Chief Scientist Alec Rose said that up until recently, innovations in the metal detection market were focused on finding smaller and smaller pieces of metal.

“Now everybody’s got this in their pocket,” Rose said, holding up his cell phone. “This didn’t exist when metal detectors were first created.”

Evolv now has over 3,000 Express systems deployed around the country, said Lemond. Typically they are sold as four-year subscriptions that can cost as much as $10.7 million dollars, which is what Prince William County Public Schools will pay for its 81 systems at its middle schools and high schools. The school division authorized the contract on May 3, and the technology will be implemented in mid-September, said division spokesperson Meghan Silas.

“Our families deserve to know that whatever innovations are out there [that are] available and at our disposal to protect their students, that we’re taking full advantage of that,” said Prince William County Superintendent LaTanya D. McDade at the school board’s April 19 meeting. “Right now [Evolv] is best in class in terms of screening, and we’re seeing more and more school divisions following suit.”

Others are not sold on Evolv’s promises. IPVM, a security and surveillance research group, has criticized Evolv for what it alleges are misleading and deceptive marketing practices. Researcher Nikita Ermolaev said even the term “weapons detection” is not entirely accurate, as it implies AI technology can find any weapon.

In 2022, a student at Proctor High School in Utica was able to get a hunting-style knife past an Evolv system and used it to stab another teenager. Earlier this year, five different law firms said they would be exploring possible lawsuits against Evolv over claims the company lied to investors about its technology’s capabilities.

“You can actually see how the marketing has changed with increased scrutiny of Evolv because historically they were claiming that their Evolv Express would create ‘weapons-free’ zones,” Ermolaev said. “Then, they switched it to ‘safe’ zones. Now, we’re in the third iteration of ‘safer’ zones.”

Tom McDermott, a “K-12 Technical Evangelist” with CEIA USA, said his company has not faced the same kind of critique as Evolv because it has built its reputation on the accuracy of its more traditional metal detection systems. In operation since 1962, CEIA has sold its systems to major league sports stadiums, maximum-security prisons, and companies like Apple and Amazon.

The company’s Opengate product, which was recently purchased by Williamsburg-James City County Public Schools in Hampton Roads through a reseller, allows the screening of similar numbers of people as Evolv’s Express system but comes at a more affordable price tag. In June, the WJCC school board approved the spending of $109,712 for six Opengate systems as well as six handheld detectors.

Despite the difference in cost between the two systems, WJCC Deputy Superintendent Daniel Keever said that finances were not part of the decision. The purchase of the systems was a result of a “threat of vulnerability” assessment conducted by the school division in the spring of 2022, and Keever said the systems won’t be used daily but on a case-by-case basis.

“We feel like our students feel safe when reporting to school and arriving at school,” Keever said. “We want to create an environment that’s warm and welcoming and a system like the one that we purchased [we] think doesn’t raise too much of an alarm.”

As far as being able to tell a gun from a cell phone, McDermott said that CEIA doesn’t need artificial intelligence to make that distinction thanks to the precision of its sensors and the immensity of its database cataloging the compositions of various objects.

“We know actually what the metal looks like in your phone compared to the barrel of a gun as an example,” McDermott said. “We already have that knowledge. We already know where the threats are, what kind of threats, and then it’s really up to the customer to decide how small a threat you want us to look for.”

Evolv isn’t the only security company capitalizing on AI. Virginia-based Omnilert offers software that works with existing security camera set-ups to spot armed individuals approaching school buildings.

Omnilert Vice President of Marketing Mark Franken said school security approaches that rely on a checkpoint method don’t cover the areas outside the school building, despite the finding by the K-12 School Shooting Database that 70% of school shootings have started outside school doors.

AI technology is new and, by nature, evolves over time, said Franken, citing an example of how workers at an automotive manufacturing company triggered Omnilert’s system because they were walking around holding large torque wrenches like firearms.

“The market is so young that the current solution is still going to be growing a lot over the next several years,” Franken said.

David Riedman, founder of the K-12 School Shooting Database, said the security checkpoint system for schools is “fundamentally flawed regardless of what technology you’re using.”

“You need lots of people, you need lots of equipment, and you need a rapid sequence of procedures that can happen to isolate somebody when you detect contraband or detect a weapon, and schools really don’t have the resources or training or staff to implement any of these systems,” Riedman said. “What’s the purpose of detection if you can’t rapidly do something about it?”