The company behind an “AI gun detection” system is being sued by the survivor of a Tennessee high school shooting after it failed to detect the handgun used by the shooter.

The lawsuit, filed last month and spotted by Ars Technica, targets Omnilert, which designed and marketed the AI weapon detection system, and System Integrations, the company that installed and maintained it. It was filed by Antonyous Henin, who was grazed in the arm during the January 2025 shooting at Antioch High School in Nashville. The perpetrator opened fire in the cafeteria, killing one student before dying of a self-inflicted gunshot wound.

Omnilert knew or should’ve known about “significant operational limitations in its gun detection systems that could result in detection failures during actual emergencies,” the suit alleges. It goes on to list numerous bold claims that Omnilert made in its marketing copy that its product apparently failed to live up to, which include boasting of “unparalleled reliability” and being able to detect a weapon “before a shot is fired.”

“Omnilert further represented that AI-powered visual gun detection ‘could have mitigated or prevented tragedy at Marjory Stoneman Douglas High School’ by identifying threats earlier — invoking one of the nation’s most devastating school shootings to convey that its product would prevent similar tragedies,” the suit continues.

The AI system was deployed after Metropolitan Nashville Public Schools approved a $1 million contract with Omnilert and System Integrations in 2023. In the aftermath of the shooting, Omnilert CEO Dave Fraser seemed to suggest that the AI detection system hadn’t failed; the shooter was merely serendipitously positioned relative to the cameras.

“The location of the shooter and the firearm meant that the weapon was not visible,” Fraser told NBC News at the time. “This is not a case of the firearm not being recognized by the system.”

An MNPS spokesperson, meanwhile, reiterated Fraser’s reasoning, by more or less invoking the adage that nothing’s perfect. “It does work, but it’s not going to work in every instance, in every spot, based on where that weapon might be visible,” the spokesperson told NBC News.

Chris Smith, one of the plaintiff’s attorneys, was skeptical of AI gun detection and said it was questionable to list a specific set of situational conditions that have to be met for the system to work as intended.

“I just thought that it was kind of bullshit. I have a Tesla, and I think Tesla’s self-driving is bullshit,” Smith told Ars. “It’s not ready for prime time! How could you possibly be entrusting of that? That’s your plan to protect kids from school shootings? Why is this any better than a metal detector?”

This isn’t the first time Omnilert has made headlines for the wrong reasons. In October 2025, its AI gun detection tech mistook a bag of Dorito chips for a weapon and brought an army of cops down on the unfortunate 16-year-old student whose only crime was toting a snack. Months later, another AI surveillance system, ZeroEyes, sent a school in lockdown after similarly hallucinating a weapon, this time fooled by a middle schooler’s clarinet.

But despite the dubious track record of these tools, AI surveillance is being increasingly used in schools across the country. Whether they represent a genuine attempt to keeps schools safer, or are the work of opportunistic startups providing a facade of safety by awing concerned parents and administrators with flashy AI promises, critics think the millions of dollars spent on deploying them would be better used elsewhere.

“I’ve never seen a school shooting where there was a lack of notification,” David Riedman, an education and security expert who maintains the K-12 School Shooting Database,” told Ars. The money spent deploying Omnilert “could have gone to a counselor or something else to a kid in crisis,” he added. “Every decision that you make is pointing away resources from something else.”

More on AI: US Soccer Scanning Videos of Millions of Youth Players to Identify New Stars