A startup has emerged that uses AI to quickly identify individuals who are about to commit mass shootings.
In the gun-owning society of America, mass shootings occur frequently. In March 2026, a shooting in downtown Austin, Texas
The Texas startup trying to stop mass shootings with AI | Mashable
https://mashable.com/article/angel-protection-ai-mass-shooter-detection-sxsw-2026
Angel Protection is an AI startup founded in 2022 in response to the Rob Elementary School shooting in Yuvalde, Texas. In that incident, a man broke into the elementary school and opened fire, resulting in the deaths of 19 children and 2 teachers.
Lewis Matthews, the founder and a father of two, is a data scientist who believes that AI and image recognition technology can be used to mitigate the damage from mass shootings and save lives. Matthews told the web media outlet Mashable, 'The problem of mass shootings is getting worse at every level, from local to federal.'
Angel Protection's technology works in conjunction with existing surveillance cameras, scanning the footage to identify individuals carrying firearms. When the system detects a firearm, it alerts human reviewers at a monitoring center and notifies law enforcement, with Angel Protection claiming it can report the incident within as little as 10 seconds of the suspect uncovering the weapon.
Angel Protection showcased its firearm detection technology at SXSW , a technology, film, and music event held in March 2026. A video of the event, filmed at SXSW, can be seen in the following X post.
Can AI be used to identify mass shooters and stop them more quickly?
— Timothy Beck Werth (@beck_werth) March 16, 2026
At SXSW I talked to Lewis Matthews, founder of Angel Protection Systems, which is trying to do exactly that. He says they can identify guns and send law enforcement photos of suspects before anyone calls 911. pic.twitter.com/L2823jDcye
The man in the camera footage is holding a model of a firearm.
If you look closely, you'll notice that the firearm is surrounded by a red border and tagged with 'rifle'.
Next, the man pulled out a handgun.
The handgun was immediately identified, surrounded by a red frame, and tagged as 'gun.'
Even when you move your firearm, you're being tracked closely.
In developing the system, Matthews focused on 'the time between a mass shooting and the reporting of the incident.' After analyzing all mass shootings that occurred between 1999 and 2025, Matthews found that the average time between the perpetrator firing the first shot and someone calling emergency services was 90 seconds.
Even when people hear gunshots, they often mistake them for more realistic noises like fireworks or doors closing, and therefore don't report it immediately. This creates a delay between the incident occurring and the police being notified. Furthermore, even when law enforcement is notified, the description of the perpetrator's appearance can vary depending on the caller, making it difficult to obtain accurate information before the police arrive.
In contrast, Angel Protection's system not only uses surveillance camera footage to quickly identify potential mass shooters, but can also directly transmit the perpetrator's facial image and location information. 'We use AI to detect guns, and then have humans verify them, reducing a task that used to take 90 seconds to less than 10 seconds. This is all done under an extremely strict system of accountability,' Matthews commented.
Privacy experts warn of the risk of AI being used for mass surveillance, but Angel Protection's chief technology officer insists that all information processing is done internally within the system to protect privacy, and that Angel Protection is only notified if a threat is identified.
According to Matthews, while involving human reviewers reduces false positives, false positives still occur in areas where carrying guns is permitted, such as Texas. One of the initial problems cited is students in the Reserve Officers Training Course carrying rifles without wearing the proper uniform.
Angel Protection is a newly established startup, but it has already checked footage from approximately 2,500 surveillance cameras in Texas, scanning them at a rate of twice per second. So far, it has partnered with schools, hospitals, and government agencies.
Related Posts:
