Chicago startup tests app that turns smartphones into gunshot sensors
SplitSec, a Chicago startup, is testing an app that uses Android smartphone microphones and on-device artificial intelligence to detect gunfire, the company and local reporting said. The company said the approach could reduce the need for fixed sensors that cities have used in the past.
SplitSec, founded in Chicago in 2025, developed the app to run entirely on users’ phones, founder and CEO Sri Deivasigamani told Fox News Digital. “We designed and built an AI model that runs entirely on your phone,” he said, adding the app is designed to “wake up only on gunshot sounds,” similar to voice assistants.
Block Club Chicago reported that SplitSec is running two no-cost trials in the 21st and 42nd wards, with discussions under way about expanding to the 5th, 29th and 38th wards. Deivasigamani told Fox News Digital that participants have reported useful feedback, including false positives on devices with sensitive microphones and increased battery drain that engineers are working to reduce.
Deivasigamani said the app detected gunfire on July 27 and 28 and that users who heard shots confirmed the alerts, though the company was not independently able to verify the incidents, he told Fox News Digital. If users confirm a gunshot, SplitSec can send an alert to people within about a one-mile radius and, when two or more users confirm the same event, triangulate a likely location, the company said.
Chicago ended its contract with SoundThinking, formerly ShotSpotter, in 2024, ending the police department’s receipt of alerts from citywide fixed sensors. SplitSec is not currently working with the Chicago Police Department or the city’s Office of Emergency Management and Communications, but Deivasigamani said the company could provide law enforcement the same anonymized map of reported gunfire if a partnership were formed.
The use of residents’ smartphones for detection has raised privacy concerns. Deivasigamani told Fox News Digital that audio analysis happens on the device and that audio clips are sent to the company only if a user reports a false positive and consents to sharing the clip for model training. “We really believe such an app needs to earn people’s trust and needs to be built in such a way that it protects privacy,” he said.
Source: Original Article
