This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
In October 2023, a Cruise autonomous robotaxi in San Francisco dragged a pedestrian 20 feet after she had been struck by a separate human-driven vehicle. The Cruise vehicle, which had come to a stop on top of the victim, attempted to pull over while she was still trapped underneath. The incident resulted in severe injuries, a multi-million dollar settlement, the suspension of Cruise's driverless permits, and federal investigations into the company's safety disclosures and autonomous systems.
Cruise has settled for between $8 million and $12 million with a pedestrian dragged by one of its autonomous vehicles in October 2023. The incident, where the pedestrian was initially hit by a human-driven car and then dragged 20 feet by the Cruise vehicle, led to the suspension of Cruise's operations and increased regulatory scrutiny.
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
Cruise failed to disclose complete footage of the incident to regulators, leading to permit suspension and federal investigations.
The Cruise robotaxi failed to perform safely under complex real-world conditions when its system failed to detect a trapped pedestrian and attempted to pull over, dragging her 20 feet.
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 severe dragging harm was caused by the Cruise AI system's decision to pull over while the pedestrian was trapped underneath, which was an unintended outcome of its programming.
Added categories peak at 2: Minor (Psychological Harm, Inferred)
An autonomous Cruise robotaxi dragged a pedestrian in San Francisco after a human hit-and-run, leading to severe injuries, regulatory suspensions, and significant corporate financial impacts. The incident represents a minor national security concern, primarily highlighting localized safety and regulatory challenges in autonomous vehicle deployment rather than systemic threats to critical infrastructure or national stability.
The Cruise robotaxi operates independently on public roads, responding to environmental inputs and making real-time driving decisions without human intervention.
High Risk: The report describes an autonomous vehicle operating in public transit and road infrastructure, which is a safety-critical application with significant implications for public safety.
The AI system is designed to operate a vehicle autonomously on public roads to transport passengers.
The incident is documented across multiple highly reputable news sources with consistent facts regarding the sequence of events, the injuries, the regulatory response, and the financial settlement.
One could attribute the primary harm to the initial human hit-and-run driver, but the robotaxi's subsequent dragging maneuver represents a distinct and severe failure of the AI system.
Technology
Flawed Post-Collision Pull-Over
The vehicle attempted to pull over while pedestrian was trapped underneath.
Inability to Stop in Time
Autonomous system could not avoid striking pedestrian thrown into its path.
Data Inputs
Incomplete Sensor Perception
System failed to detect pedestrian trapped under the vehicle after collision.
Human Factors
Hit-and-Run Human Driver
A human driver struck the pedestrian initially and fled the scene.
Process and Methods
Flawed Accident-Response Logic
The system prioritized clearing the lane over pedestrian safety.
Incomplete Video Sharing
Cruise failed to initially show complete footage to the California DMV.
Management
Aggressive Deployment Push
Management prioritized rapid scaling and driverless deployment over safety.
Inadequate Risk Assessment
Management failed to assess risks of post-collision maneuver behaviors.
Regulatory Environment
Permits for Driverless Deployment
Regulators allowed driverless testing before validating post-collision safety.
AI system involved
Identified by automated extraction; can differ from the alleged developers above.
