Traumatized teens sue school, cops after Doritos bag triggers AI gun alert
First reported by Ars Technica ·
AI gun detection systems in schools now carry the risk of triggering disproportionate police responses to everyday objects.
Four Black high school students in Baltimore County were subjected to a traumatic police response after an AI gun detection system, Omnilert, falsely identified a bag of Doritos as a firearm in October 2025. School officials alerted police, who arrived with guns drawn, ordered the teens to the ground, handcuffed them, and searched them, despite school officials having already determined there was no threat. The students, who were traumatized by the incident, are suing Baltimore County Public Schools (BCPS) and the police department, alleging violations of their constitutional rights and seeking an injunction to prevent the use of Omnilert without proper protocols. The lawsuit highlights that over 99% of BCPS's AI gun alerts are false positives. An Inspector General report found no bias in the AI but acknowledged human and procedural errors, recommending improved training and alert review processes.
The incident underscores the significant risks associated with the widespread deployment of AI in public safety, particularly in educational settings where the stakes for misidentification are exceedingly high. The lawsuit alleges that a combination of flawed technology and inadequate human oversight led to a severe overreaction, impacting students' constitutional rights and mental well-being. This case brings to light the critical need for robust verification protocols and human review processes before AI-generated alerts result in potentially dangerous law enforcement actions, especially when the technology is marketed with claims of "near-zero false positives."
The lawsuit against BCPS and Omnilert suggests a broader market concern about the reliability and implementation of AI security solutions. It raises questions about the responsibility of technology vendors in ensuring their products are not misused or prone to triggering unwarranted panic. Future deployments of such systems may face increased scrutiny and demand for transparency regarding their accuracy rates and the necessity of comprehensive training for human operators to prevent traumatic and potentially illegal encounters.
AI-written summary. May contain errors.