UMPD has spent more than $135,000 on AI gunshot detection with payments spanning 2017 to 2023, despite concerns about the technology’s accuracy, according to records obtained by The Diamondback.
University of Maryland Police use ShotSpotter — an artificial intelligence-based acoustic gunshot detection system owned by the technology company SoundThinking — to alert campus police in the event of an active shooter. But experts and activists urge caution when deploying this technology due to concerns about false positives and public safety.
When police receive a notification from ShotSpotter, they are dispatched to look for evidence of a gunshot, according to UMPD spokesperson Lt. Rosanne Hoaas.
“This can range from officers checking with any potential witnesses who might be in the area, to phone calls UMPD may receive,” Hoaas wrote in a statement to The Diamondback. “After the incident is cleared, we notify SoundThinking what our investigation revealed.”
University police partnered with SoundThinking around 2015 and chose McKeldin Mall for the first sensor deployment because of the location’s high foot traffic and easy accessibility for potential threats, UMPD police chief David Mitchell said.
According to the university’s general counsel office’s response to a Diamondback public information request, ShotSpotter “has no confirmed detections” at this university from 2017 to 2024.
Jonathan Manes, senior counsel at Northwestern Law School’s MacArthur Justice Center who researches ShotSpotter in Chicago, Illinois, said the technology often leads to no traces of gunfire.
“We found that about 90 percent of the time [in Chicago], police found no indication of gunfire whatsoever,” Manes said. “The system’s never actually tested to see how frequently or how easily it can be triggered in response to those kinds of non-gunfire, innocuous noises.”
[Here are 3 takeaways from UMPD’s 2023 Annual Security Report]
University of Maryland Police Chief David Mitchell told The Diamondback that a shuttle bus that drove over a water bottle once triggered a ShotSpotter alert at this university.
These false alerts can present dangerous legal and public safety implications, Manes said, since police often approach the scene assuming a shooter is nearby.
The Diamondback filed multiple public information requests to verify the dates and locations where the ShotSpotter system was triggered on this campus.
This university denied a series of public information requests on the basis that the information would reveal investigative and surveillance techniques and “jeopardize the public security of buildings, structures, or facilities, or endanger life,” an April response from the university’s general counsel office read.
SoundThinking wrote in an Oct. 15 statement to The Diamondback that its system has a 97 percent accuracy rate and faces financial penalties for performance with less than 90 percent accuracy.
Hoaas wrote in a statement that university police used live gunfire to calibrate and test ShotSpotter devices during Thanksgiving break in 2016 and winter break in 2017.
In a statement to The Diamondback on Oct. 15, SoundThinking stated that live gunfire was “not necessary” to calibrate its devices. On Nov. 6, the company added that it has previously used controlled live gunfire testing, but discontinued it “years ago.”
Eric Piza, a professor of criminology and criminal justice at Northeastern University, said ShotSpotter is designed to be a reactive tool meant to address gun violence after it occurs rather than preventing crime.
“Police are essentially using the ShotSpotter detections just to support that reactive function, and they aren’t doing much to use it to develop more problem-oriented solutions,” Piza said.
[UMPD reports more than 70 e-scooter thefts since Sept. 1]
Activists who oppose ShotSpotter have long stressed that the technology leads law enforcement to over-police neighborhoods that experience gun violence.
Ed Vogel, a member of the “StopShotSpotter” national coalition, said he believes ShotSpotter makes communities more violent because police approach situations expecting violence.
WIRED magazine released leaked ShotSpotter locations across the country in February, which include 12 sensors on top of academic and residential buildings at this university.
UMPD did not confirm these locations. Mitchell told The Diamondback that the number of locations is “in that vicinity.”
The Diamondback attempted to confirm ShotSpotter locations at this university through public information requests. The university also denied these requests because the information would “reveal surveillance techniques.”
Mitchell asked for The Diamondback’s discretion when reporting these locations. When asked to confirm the ShotSpotter sensor locations, Hoaas wrote in a statement that the department does not reveal “specific details when it comes to security measures, such as the exact location of acoustic sensors” as a matter of practice.
A SoundThinking spokesperson wrote in a statement to The Diamondback that “the precise locations of sensor locations represent a security and operational risk.”
But Mitchell stated in an interview with The Diamondback that the likelihood of an attacker tampering with ShotSpotter sensors is low.
“Most people who engage in active shootings I don’t think go around looking for ShotSpotters,” Mitchell said.
Other municipalities, such as Washington, D.C., release data from their ShotSpotter systems to the public.
University police are looking to expand the ShotSpotter system, Mitchell said.
Mitchell said he is interested in partnering with neighboring jurisdictions, such as Hyattsville and Riverdale Park, to maintain “situational awareness” about local gunfire incidents. He hopes to integrate the technology into other campus security features, such as security cameras, to grow the system’s capabilities.
Advocates who oppose ShotSpotter, including members of “StopShotSpotter” such as Vogel, say using surveillance technology to handle public safety concerns does not address the root causes of these issues.
“If the question is safety, we need to really rethink how we structure and use public money towards improving people’s lives, rather than investing in technologies that have no proven ability to do any of that,” Vogel said.
