This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
On July 24, 2022, a Tesla Model 3 operating on Autopilot collided with a motorcycle on Interstate 15 in Draper, Utah. The motorcyclist was killed instantly in the rear-end collision. The driver reported that they did not see the motorcyclist, and the incident has drawn scrutiny from the National Highway Traffic Safety Administration (NHTSA) regarding the ability of Tesla's driver-assist systems to detect motorcycles.
A Tesla Model 3 operating on Autopilot mode slammed into the back of a Harley-Davidson motorcycle on an interstate in Utah, throwing the rider from the bike and killing him instantly.
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
The driver relied on the Autopilot system to monitor the road in the dark, leading to a failure to intervene and prevent the fatal collision.
The Autopilot system failed to perform reliably in dark conditions, specifically failing to detect and avoid a motorcycle, which represents a critical capability limitation.
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 crash was caused by the failure of the Tesla Autopilot system to detect and brake for the motorcycle in front of it while operating in real-world conditions.
Added categories peak at 2: Minor (Psychological Harm, Inferred)
A fatal traffic accident in Utah involving a Tesla Model 3 operating on Autopilot. The incident highlights ongoing safety and regulatory challenges surrounding consumer autonomous driving technology but carries negligible national security implications.
The Autopilot system operates independently to keep the car in its lane and maintain distance from other vehicles, making real-time driving decisions based on camera inputs without explicit human planning.
High Risk: The system is an AI safety component used in road traffic and vehicle control, which has significant implications for physical safety and public interest.
The primary function of Tesla Autopilot is to assist or automate driving tasks, keeping the vehicle in its lane and maintaining safe distances.
The reports provide clear, consistent, and detailed factual accounts of the specific crash in Draper, Utah, including the date, location, vehicle models, victim identity, and the driver's explicit statement regarding Autopilot use.
The crash could be interpreted primarily as driver negligence, but the driver's reliance on Autopilot and the system's failure to brake support the AI safety classification.
Technology
Camera-Only Vision System
Removal of radar reduced detection capability in dark conditions.
Inadequate Target Classification
Autopilot system failed to recognize and stop for a motorcycle ahead.
Poor Low-Light Performance
The camera-based system struggled to detect objects in nighttime conditions.
Data Inputs
Lack of Radar Sensor Data
Exclusion of radar sensor inputs left only camera data for object detection.
Nighttime Visual Constraints
Darkness limited the visual data inputs captured by the vehicle cameras.
Human Factors
Driver Inattention
The driver did not see the motorcyclist and relied too heavily on Autopilot.
False Sense of Security
Marketing and design lured the driver into overestimating system capabilities.
Process and Methods
Public Beta Testing
Testing unfinished software on public roads with untrained customer drivers.
Inadequate Safety Verification
Insufficient validation of Autopilot safety before deployment to public.
Management
Prioritizing Rapid Deployment
Management pushed Autopilot updates quickly despite known detection risks.
Removal of Redundant Sensors
Decision to eliminate radar in favor of a camera-only vision approach.
Regulatory Environment
Absence of Federal Regulations
No federal standards govern partially automated driver-assist systems.
Delayed Regulatory Intervention
Regulators investigated multiple crashes without enforcing timely recalls.
AI system involved
Identified by automated extraction; can differ from the alleged developers above.
