This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
A ZeroEyes AI surveillance system at a Florida middle school triggered a school lockdown and police response after misidentifying a student's clarinet as a rifle. The student was participating in a themed dress-up day. While the company maintains the alert was a proactive safety measure, the incident highlights the risks of false positives in AI-driven threat detection systems in schools.
An AI-powered ZeroEyes surveillance system deployed at a Florida middle school reportedly flagged a student's clarinet as a firearm, triggering a school lockdown and police response. Officers reportedly searched classrooms and questioned a student after the alert described a suspected weapon. No threat was reportedly found. The incident reportedly caused disruption to classes and subjected students and staff to lockdown conditions and police intervention despite the absence of an actual weapon.
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.
Also classified as
School administrators and law enforcement relied on the automated AI alert to initiate a high-stress lockdown and armed police response before fully verifying the threat.
The AI threat-detection system failed to robustly distinguish between a musical instrument and a firearm, leading to a false positive alert.
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
The incident was caused by the ZeroEyes AI system's false positive identification of a clarinet as a rifle after it was deployed in the school.
Added categories peak at 2: Minor (Psychological Harm, Direct)
An AI threat-detection system misidentified a student's clarinet as a rifle, triggering a localized middle school lockdown in Florida. The incident highlights ongoing reliability challenges and potential human-bias issues in AI surveillance, but carries negligible national security implications.
The system acts as an Assistant because it flags potential threats to human reviewers who must validate them before law enforcement is alerted, thereby augmenting human surveillance capabilities rather than acting fully autonomously. The report states 'trained employees review alerts before they are sent'.
High Risk: The system is used for law enforcement and school safety threat detection, which has significant implications for public safety and fundamental rights. The report notes it is an 'AI-powered surveillance system used by Lawton Chiles Middle School' to 'alert law enforcement'.
The system's explicit design and deployment goal is to scan video surveillance footage to detect weapons and security threats.
The report provides a clear, detailed account of the specific false positive event, including the perspectives of the school, the AI developer, and external safety experts. The role of the AI is explicitly defined, and the sequence of events is well-documented.
None
Technology
Object Misclassification
The AI system mistakenly flagged a clarinet as a shouldered rifle.
Camera Distance Limitations
AI surveillance systems can miss threats if targets are too far from cameras.
Overreliance on Visual Patterns
The system relies on visual matches of weapon-holding poses, causing errors.
Data Inputs
Deceptive Visual Posture
The student held a clarinet in a manner resembling a shouldered rifle.
Lack of Contextual Data
The system lacked situational data like the school's themed dress-up day.
Human Factors
Human Reviewer Confirmation Bias
Evaluators err on the side of warning when AI flags a potential weapon.
Student Playful Behavior
The student held the musical instrument in a military-style posture.
Process and Methods
Proactive Alerting Bias
The verification process favors dispatching alerts even with minor doubts.
Rapid Verification Pressure
Human reviewers must make split-second decisions, increasing error rates.
Management
Tolerance of False Positives
Management prefers false alarms over the risk of missing a real threat.
Regulatory Environment
Unproven Tech Deployment
AI surveillance is deployed in schools without established safety standards.
AI system involved
Identified by automated extraction; can differ from the alleged developers above.
