Gunfire Detection Technology Spreads, Raises Privacy Concerns

When a gun is fired in a city, it’s not a given that people who hear the shot will call the police. 

In February, Houston Police Chief Art Avecedo said that was the animating factor for his decision to work with ShotSpotter, a company that will deploy a gunfire detection system across five square miles of his city. 

“We want to know about the gun violence that’s going on …. that we’re not being called on,” he said at a news conference. “We don’t want to wait until somebody gets struck by the gunfire … to actually be aware of the gunfire. Because that will help us deploy in a proactive manner on the front end.”

But the technology has also provoked debates in several cities about tradeoffs between surveillance and security, with people living in the majority black and Latino communities where it is often deployed sometimes questioning if they really see a benefit. Civil rights advocates have also raised concerns about whether the technology could pave the way for more intrusive applications to monitor conversations. 

ShotSpotter’s system—audio sensors equipped with artificial intelligence systems trained to identify the sound of gunshots—aims to give police a relatively precise location of gunfire. When three sensors pick up the same sound, the system triangulates the source and alerts the company’s Incident Review Center in Newark, California, where human analysts confirm that the sound came from a gun and notify local police. According to the company, the entire process usually happens within 45 seconds of the trigger pull. 

At least 96 cities have Shotspotter technology monitoring for gunshots in their streets (while Shotspotter is not the only system capable of detecting gunshots, the systems sold to city and county governments are almost exclusively provided by Shotspotter). Law enforcement in many of those cities have praised the technology as instrumental in their ability to respond to gun violence incidents. In Cincinnati, Police Lt. Col. Paul Neudigate told local news station WCPO in 2018 that "prior to the implementation, we were only responding to 15% of the gun incidents. We're responding to 100% of them now.” 

This month, Neudigate said Cincinnati would expand its network into new neighborhoods, with ShotSpotter equipment eventually installed in 13 square miles at a cost of $900,000 annually. But he emphasized it is worth it, saying the cost of treating shooting victims is much higher.

A 2016 study that analyzed ShotSpotter data against calls to 911 found that “gunshots that do not hit anyone are often not reported to police, and this selective underreporting may be particularly problematic in the most violent neighborhoods if residents don’t trust the police to be helpful.” The research conducted in Washington, D.C. and Oakland determined that an average of only 12% of gunfire incidents resulted in a 911 call.

Advocates for gunfire detection systems say that responses to shooting incidents are not only more efficient when monitoring technologies are deployed, but that those quick responses are more likely to result in recovered forensic evidence like shell casings. 

But whether this more active follow-up by police leads to results is still up for debate. A 2016 investigation from the Center for Investigative Reporting that tracked more than 3,000 ShotSpotter alerts over two-and-a-half years in San Francisco found that only two of those resulted in an arrest. 

Others have raised concerns about the percentage of alerts that appear to be false alarms. A review by WNYC into Newark, New Jersey’s ShotSpotter alerts found that 75% of them were false alarms between 2010 and 2013, but police were deployed anyway in case there was an active shooter. In that city, the system led to 17 arrests, the radio station found.

ShotSpotter did not respond to requests for comment. During a February earnings call about 2019 performance, Ralph Clark, the company’s president and CEO, touted the company’s expansions into new cities in Puerto Rico, as well as in Dayton, Ohio. He also noted the Houston rollout, which he said included the company offering a free trial that would turn into a subscription if the police department was pleased with the system.   

But not all cities that have considered the system decide to move forward. In Durham, North Carolina, council members evaluated a proposal by the police department last year to deploy ShotSpotter as part of a larger investment, but ultimately rejected it. Council member Vernetta Alston questioned the effectiveness of the system, while also saying she didn’t want particular “communities to feel targeted by geofencing,” according to the News & Observer. 

Some privacy experts worry that the company potentially has too much power to record daily life. Jay Stanley, a senior policy analyst with the ACLU Speech, Privacy, and Technology Project, said that microphones in public places are always a concern. “We’re subject to so much video recording, but audio is potentially more invasive,” Stanley said. 

ShotSpotter’s privacy policy states that the technology was developed with an eye toward “respecting individual privacy” by limiting the capacity for audio surveillance. “It is important to note that human voices and street noise will never trigger a sensor because they do not produce an instantaneous, sharp sound,” the website reads. “Live streaming of sensor audio is not possible by company employees, police or third parties.”

An independent evaluation of the technology by New York University’s Policing Project agreed that the risk of voice surveillance was low. Recommendations from that report were integrated into ShotSpotter’s system, including a reduction in the duration of audio stored on sensors, a commitment to denying requests and subpoenas for sensor audio, and improved internal controls regarding audio access.

When Clark did an interview with Stanley in 2015, he said that fears of surveillance were overblown. “If you’re really worried about that, what about your cell phone?” he asked. “If you’re worried about NSA boogeymen, they’re not going to be using our sensors, they’ll be using your phone. It’s in your pocket and has a better microphone.”

Stanley said he “firmly rejects” that argument. “Just because our privacy is being invaded through techniques A, B, and C, doesn’t mean we stop fighting invasion through techniques D, E, and F,” he said. “Cell phones remain an enormous privacy threat, but if we want to create a society that reflects our values, we have to keep working on multiple fronts.”

Stanley said that he also has concerns around differential deployment, which could lead to increased police presence in predominantly black and Latino neighborhoods. Some city officials, like those in Chicago, have recognized this concern and started giving neighborhood advocates tours of the police technology control centers to build trust and alleviate fears of surveillance. Some residents in neighborhoods that both majority black and the site of frequent gunfire, like Englewood, have said they appreciate the new measures. Between 2017, when the audio sensors were first installed for a pilot in Englewood, and the next year, shootings were down by 52%.

Perry Gunn, executive director of the community organization Teamwork Englewood, told The New York Times in 2018 that cameras and audio sensors monitoring for gunfire have improved the feeling of neighborhood safety. “It’s working,” Gunn said. “I think it is a wonderful resource to have technology to fight crime.”

As the technology spreads to more and more cities, some local officials see it as an on-ramp to other surveillance systems. When the Cleveland City Council voted in 2019 to deploy a pilot of gunshot detection technology, Councilmember Blaine Griffin pushed the council to consider even more options. “I don’t think it stops here,” Griffin said before the council. “I think we have to look at drones and other technology.”

Privacy experts often make the same point in opposition to technologies like gunfire detection systems, arguing that they pave the way for more intrusive surveillance methods. 

“We know that always-on AI monitoring technologies are likely to spread,” said Stanley. “That doesn’t mean we never allow tools that have that capability, but we’re going to be confronting harder tools in the future. We need to be vigilant against erosions of privacy in public spaces.”