This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
An EasyMile autonomous shuttle operating in Columbus, Ohio, performed an unexpected emergency stop while traveling at 7 mph, causing a passenger to fall and sustain minor injuries. Following the incident, the National Highway Traffic Safety Administration (NHTSA) suspended EasyMile's ability to carry passengers in its US shuttle operations pending an investigation.
A self-driving shuttle deployed by Smart Columbus in Linden neighborhood unexpectedly stopped on the street, which caused a woman to fall onto the floor from her seat.
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.
The shuttle's AI system failed to perform robustly by incorrectly identifying a non-existent obstacle, leading to an unnecessary and hazardous emergency stop.
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 autonomous vehicle's AI system making an unexpected emergency stop due to a false positive detection during active deployment.
An EasyMile autonomous shuttle in Columbus, Ohio, executed an unexpected emergency stop due to a false positive sensor detection, causing minor injury to one passenger. This localized safety failure prompted a standard regulatory suspension of passenger operations by NHTSA, representing a minor physical security and critical infrastructure impact with negligible broader national security implications.
The shuttle operates independently in real-time, responding to environmental inputs and making braking decisions without explicit human planning or direct steering control.
High Risk: The AI system is deployed in public transport and critical infrastructure, where operational failures pose direct risks to physical safety.
The system's primary goal is to navigate and drive the shuttle autonomously, which includes executing emergency stops when obstacles are detected.
The reports provide consistent, clear details about the physical event, the passenger's injury, the immediate regulatory response, and the technical context of false-positive braking.
None, as the physical failure of the autonomous braking system is clearly documented as the sole cause of the passenger's fall.
Technology
Conservative Braking Algorithm
The vehicle was programmed to stop suddenly if any problem was detected.
False Positive Detection
Sensors detected a non-existent obstacle, causing a ghost braking event.
Data Inputs
Ghost Obstacle Sensor Data
Sensors registered a ghost obstacle, leading to unnecessary emergency braking.
Human Factors
Lack of Passenger Restraints
Passengers did not wear seatbelts, leaving them vulnerable to sudden stops.
Unsafe Seating Configuration
Forward-facing seats lacked barriers or belts to prevent passengers falling.
Process and Methods
Prototype Passenger Testing
Carrying passengers during the risky prototype phase of the shuttle.
Management
Premature Public Deployment
Deploying experimental autonomous shuttles for public transit use too early.
Regulatory Environment
Inadequate Regulatory Oversight
NHTSA lacked leadership in regulating advanced driving automation systems.
AI system involved
Identified by automated extraction; can differ from the alleged developers above.
