Gunshot detection

Acoustic gunshot detection uses networks of microphones to locate gunfire and alert police. It is concentrated in US cities and is one of the few police technologies with years of evidence behind it, which has led several cities to drop it.

Guide

What it is

  • Sensor networks microphones on lamp posts and buildings listen for loud, impulsive sounds and triangulate where they came from.
  • Human review a reviewer at the vendor's centre usually listens to the audio before the alert reaches police dispatch.
  • One dominant vendor SoundThinking's ShotSpotter, deployed mainly in American cities, with little presence in UK policing.

How it is used

Alerts send officers to a location, often where nobody has called the emergency services: in Denver, 87.6 per cent of alerts had no matching 911 call. The record tracks deployments, independent audits and the growing number of cities ending their contracts.

Key tensions

Alerts versus outcomes

Denver's 25,217 alerts from 2020 to mid-2026 led to 655 arrests. Supporters see unreported gunfire being found; critics see thousands of dispatches with no confirmed result.

Confirmation and bias

A New York City Comptroller audit found 87 per cent of alerts led officers to places with no confirmed shooting, and a public defender analysis found deployments fell disproportionately on Black and Latino neighbourhoods. The vendor disputes this.

Cities walking away

Chicago chose not to renew in 2024, Cambridge, Massachusetts voted to end its contract in 2026, and Denver's runs out at the end of 2026. Some want a better system rather than none.

View all gunshot detection records Latest coverage

In depth

The full guide, about 4 minutes to read.

How acoustic gunshot detection works

Acoustic gunshot detection systems use networks of sensors, typically mounted on lamp posts and buildings across a defined area of a city, to listen for loud, impulsive sounds consistent with gunfire. When several sensors pick up a matching sound within milliseconds of each other, the system triangulates a location and forwards the alert, usually after a human reviewer at the vendor's review centre has listened to the audio and judged it likely to be gunfire, to a police dispatch centre. The pitch is straightforward: many shootings are never reported by a call to emergency services, so a system that detects gunfire independently of any witness picking up a phone should get officers to a scene faster and uncover incidents that would otherwise go entirely unrecorded.

SoundThinking, the company behind the best known system, ShotSpotter, is by far the dominant vendor in this space, with deployments concentrated in American cities. The technology has comparatively little presence in UK policing, where its role in coverage on this site sits mainly as a case study in how a widely adopted surveillance technology gets evaluated, and in several prominent cases discontinued, once cities have several years of operational data to examine.

What the evidence actually shows

The core dispute about acoustic gunshot detection is not really about whether the underlying audio triangulation works. It is about what happens after an alert is generated. Denver Police Department disclosed that from 2020 through June 2026, its ShotSpotter system produced 25,217 alerts, leading to 655 arrests and the recovery of 721 firearms, with shell casings recovered as evidence in 6,672 of those alerts. An average of 87.6 per cent of alerts had no corresponding 911 call, which supporters point to as evidence the system is surfacing gunfire that would otherwise go unreported. Critics point to the same figures rather differently: several thousand alerts sending officers to a scene for every few hundred arrests represents a very high proportion of dispatches that produce no confirmed enforcement outcome at all.

New York offers the most detailed independent scrutiny available. An audit by the Office of the New York City Comptroller found that 87 per cent of the time, ShotSpotter alerts sent NYPD officers to locations where no confirmed shooting had occurred, and concluded the department substantially overstated the response time improvements the system delivered. A separate analysis by the public defender organisation Brooklyn Defenders, examining nine years of NYPD performance data, reported similarly low confirmation rates and found that ShotSpotter related police deployments fell disproportionately on Black and Latino neighbourhoods. The vendor disputes characterisations of the technology as unreliable, and has said in response to similar criticism elsewhere that its system is, in its own description, proven and unbiased.

Cities are increasingly walking away

The practical consequence of this pattern of evidence is that a growing number of American cities have chosen not to renew their contracts. Chicago announced in February 2024 that it would not renew its citywide agreement. Cambridge, Massachusetts voted in May 2026 to terminate its contract within 90 days, with councillors citing false positive rates, the cost of the programme, and concerns about the vendor's data sharing arrangements with federal authorities, including immigration enforcement. Denver's own contract, running since 2015, is set to expire at the end of 2026, with the city already reducing coverage by removing sensors from areas generating the fewest alerts while it decides whether to rebid the contract to a different vendor or discontinue the programme altogether.

None of these decisions amount to a conclusive verdict that acoustic gunshot detection cannot work as intended. Some officials in cities considering non-renewal have said explicitly that they are looking for a system that performs better, not abandoning the underlying concept. But the pattern across several major cities within a short period, after years of real operational data had accumulated, distinguishes gunshot detection from many other police AI systems that remain largely unevaluated in practice. This is one of the few areas of police technology where a reasonably large sample of before and after evidence exists, and a meaningful number of buyers have concluded, on the basis of that evidence, that the return did not justify the cost or the surveillance footprint.

What is not settled

There is no independent, peer reviewed study establishing a reliable figure for how often acoustic sensors correctly detect actual gunfire, as distinct from how often a human reviewer subsequently confirms an alert as plausible gunfire, which is a different and lower bar. Nor is it settled how far the disproportionate placement of sensors in specific neighbourhoods, a pattern raised in several of the audits and reports above, reflects genuine data driven targeting of high crime areas as vendors and departments maintain, or reflects and reinforces existing patterns of over policing, as critics argue. As more cities allow contracts to lapse, whether the vendor's product genuinely improves, whether departments simply migrate to a different acoustic detection provider, or whether the technology's American footprint contracts significantly over the coming years, remains to be seen.

Records on the tracker

Open the full tracker

17 records on gunshot detection, most recently checked first: 15 in use now, 2 withdrawn and 0 on the horizon. Each is one police force or public body using one system.

  1. United StatesInvariant Corporation2025

    US Customs and Border Protection — Acoustic gunshot detection pilot (Invariant Corporation)

    Southwest border

    TrialChecked
  2. United States

    Drug Enforcement Administration — Gunshot detection system

    Nationwide

    OperationalChecked
  3. United StatesSoundThinking

    Bureau of Alcohol, Tobacco, Firearms and Explosives — Access to ShotSpotter via external state and local systems

    Nationwide

    OperationalChecked
  4. United StatesFlock Safety

    Richmond Police Department (Virginia) — Flock Safety gunshot detection

    Richmond, Virginia

    OperationalChecked
  5. United StatesSoundThinking (ShotSpotter)

    Columbia Police Department (South Carolina) — ShotSpotter acoustic gunshot detection

    Columbia, South Carolina

    OperationalChecked
  6. United StatesSoundThinking (ShotSpotter)2015

    New York City Police Department — ShotSpotter acoustic gunshot detection

    New York City

    OperationalChecked
  7. United StatesSoundThinking (ShotSpotter)2014

    Pittsburgh Bureau of Police — ShotSpotter acoustic gunshot detection

    Pittsburgh, Pennsylvania

    OperationalChecked
  8. United StatesSoundThinking (ShotSpotter)

    Cambridge Police Department — ShotSpotter acoustic gunshot detection

    Cambridge, Massachusetts

    OperationalChecked
  9. South AfricaSoundThinking (formerly ShotSpotter)2016

    City of Cape Town — ShotSpotter acoustic gunshot detection

    Hanover Park, Manenberg and Lavender Hill, Cape Town

    DiscontinuedChecked
  10. United StatesSoundThinking (ShotSpotter)2015

    Denver Police Department — AI Gunshot Detection

    Colorado

    OperationalChecked
  11. United StatesSoundThinking (ShotSpotter)2020

    Albuquerque Police Department — AI Gunshot Detection

    New Mexico

    OperationalChecked
  12. United StatesSoundThinking (ShotSpotter)2014

    Worcester Police Department — AI Gunshot Detection

    Massachusetts

    OperationalChecked
  13. United StatesSoundThinking (ShotSpotter)2022

    Richland County Sheriff's Department — AI Gunshot Detection

    South Carolina

    OperationalChecked
  14. United StatesSoundThinking (ShotSpotter)2022

    Macon-Bibb County Sheriff's Office — AI Gunshot Detection

    Georgia

    OperationalChecked
  15. United StatesSoundThinking (ShotSpotter)2013

    Peoria Police Department — AI Gunshot Detection

    Illinois

    OperationalChecked
  16. United StatesSoundThinking (ShotSpotter)

    City of Fayetteville (North Carolina) — AI Gunshot Detection

    North Carolina

    OperationalChecked
  17. United StatesSoundThinking (ShotSpotter)

    Chicago Police Department — AI Gunshot Detection

    Illinois

    DiscontinuedChecked

Latest coverage