Part of this story originally appeared as a segment in Battleground Baltimore on Oct. 29, 2021.
Baltimore City’s Board of Estimates voted to approve an additional $759,500 to continue the usage of ShotSpotter, gunshot detection technology that uses artificial intelligence to identify gunshots and reports the location of those gunshots.
The vote on extending the city contract with ShotSpotter surveillance technology was deferred last week by the Board of Estimates, after City Councilperson Ryan Dorsey and activists such as DeRay Mckesson challenged the technology’s efficacy and called attention to concerns about its surveillance capabilities.
Mckesson spoke out against the renewed contract last week, tweeting that ShotSpotter “is not a crime-fighting tool, has no scientific validity, & is not a value-add,” and encouraged Mayor Brandon Scott and Comptroller Bill Henry, who are on the Board of Estimates, to defer the vote.
“On their website, @ShotSpotter claims that they are responsible for a 15% reduction in shootings from the prior year. What’s their source? It’s the BPD Crime Reduction Plan that does not support this claim at all. Again, we don’t need ShotSpotter,” Mckesson tweeted. “By delaying the vote, you’ll be able to see more data that shows that @ShotSpotter is a liability in communities and not a value-add. Their tool is *not* a crime-fighting tool and, again, has not been proven to hold up their claims at all. Please delay the vote.”
The vote was deferred by one week. Councilperson Ryan Dorsey criticized the technology after last week’s deferral: “Shotspotter is a joke. It has delivered no better outcomes for Baltimore, and will continue to do so for as long as we waste money on it,” Dorsey tweeted.
ShotSpotter has been in use in Baltimore since 2018 and the approval of the $759,000 to extend the contracts put the total amount of money spent by the city on ShotSpotter at a little over $3 million.
The technology has been challenged because it has the capability to record people’s conversations, and, as Vice reported earlier this year, it has been alleged that ShotSpotter has changed its AI-generated results. In a shooting case in Chicago, Vice reported that ShotSpotter analysts, at the request of the police, changed a noise ShotSpotter categorized as fireworks to gunshots. Months later, Vice reported, the ShotSpotter changed the location of the supposed gunshots—which were initially categorized as fireworks, Vice reported—to put those “gunshots” closer to a suspect’s car.
ShotSpotter is currently suing Vice for defamation for their reporting, “Police Are Telling ShotSpotter to Alter Evidence From Gunshot-Detecting AI.”
“ShotSpotter categorically denies any allegations that we manipulate any details of an incident at the request of the police,” a spokesperson for ShotSpotter told Battleground Baltimore over email.
After a week of deliberation, the technology was unanimously approved by the city’s spending board.
Mayor Scott, who very vocally opposed the use of an evidence-gathering surveillance plane last year, said he is “the biggest skeptic” of ShotSpotter, but approved it anyway.
“I am the biggest skeptic of ShotSpotter,” Scott said. “But saying that and also knowing that this is the final renewal for this contract and, as I have already directed CitiStat to do an in-depth analysis of this tool… This really for me is about getting to people who are the victims of gun violence who no one is calling for.”
Scott explained that he intends to use the gunshot location data generated by ShotSpotter for the Mayor’s Office of Neighborhood Safety and Engagement. The data, Scott said, can be used to figure out which communities in Baltimore are enduring the most gunfire.
“We have the opportunity to leverage the technology in a multi-agency, multi-pronged way,” Shantay Jackson, director of the Mayor’s Office of Neighborhood Safety and Engagement, said.
Baltimore City Council President Nick Mosby said he is “very supportive” of ShotSpotter being used “in the crime fight.”
City Comptroller Henry alluded to some of the concerns about the technology’s efficiency, saying that he “follows the conceptual” about ShotSpotter’s potential, but hoped it can recover more evidence that can be used to prosecute shootings and homicides.
The Chicago Office of Inspector General, for example, released a report earlier this year that showed that an examination of 50,000 alerts from ShotSpotter in 2020 resulted in evidence of a gun-related crime just 9% of the time.
“I actually wasn’t a skeptic of the program when it was proposed, I thought it was an intriguing prospect, but I wanted to actually see some metrics,” Henry said. “I wanted to see some results.”
During today’s Board of Estimates meeting, the Baltimore Police made the case for ShotSpotter.
“This is an important investigative tool for the [police department]. It’s been around since 2018,” Eric Melancon, chief of staff for the Baltimore Police Commissioner, argued. “It uses a series of audio sensors to alert our officers of gunfire. That’s improved our response time to these incidents.”
Lieutenant Colonel John Herzog of the Baltimore Police Department stressed that 88% of the detections by ShotSpotter had no corresponding 911 call. The implication by Herzog and others in favor of ShotSpotter is that the technology captures significant amounts of gunfire that is not being called in to 911 by residents.
What was not explained by police, however, is how many of those detections were actually gun fire rather than, for example, fireworks or cars backfiring—two loud city noises that ShotSpotter regularly detects.
McKesson noted that the 88% number was intentionally deceiving. “This is wildly disappointing. And doesn’t bode well for a deliberative bodies [sic] that pretends to believe in data,” he tweeted after the funding was approved. “During the presentation today, the @BaltimorePolice openly misled the BOE and all of their members just sat and accepted it. We deserve better.”