ShotSpotter's gunshot-detection AI system triggered a police response that led to a DUI arrest, but the company refused to provide documentation about the system's accuracy and analyst review processes when subpoenaed by defense attorneys.
ShotSpotter operates a gunshot-detection technology system using acoustic sensors that identify gunshots and trigger police alerts. On November 7, 2021, a ShotSpotter alert was recorded near Hamlin and Lake Street in Chicago's Garfield Park area, prompting at least eight Chicago police officers to respond. Two officers arrested a defendant a few blocks away on Central Park Avenue for DUI. The Cook County Public Defender Office subpoenaed ShotSpotter as a third party to obtain documents about the alert that led to the arrest, including analyst qualifications, training materials, instances of alert reclassification, methods for reclassification, data on sensor misidentification of gunfire, and data on missed gunshots. Instead of complying with the subpoena, ShotSpotter's attorneys requested the judge hold the company in 'friendly' contempt of court to allow them to appeal the discovery order. The sensors can incorrectly identify fireworks and other noises as gunshots, and the company employs analysts who review alerts in real time and can reclassify them retroactively. A separate civil lawsuit filed by Lucy Parsons Labs cited cases where ShotSpotter alerts led to arrests, including a 65-year-old grandfather held in jail for nearly a year before his case was dismissed due to insufficient evidence.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
AI systems that fail to perform reliably or effectively under varying conditions, exposing them to errors and failures that can have significant consequences, especially in critical applications or areas that require moral reasoning.
AI system
Due to a decision or action made by an AI system
Unintentional
Due to an unexpected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed