This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
On August 14, 2023, two Cruise autonomous vehicles were reported to have blocked an ambulance transporting a critically injured patient in San Francisco. While initial reports and internal Fire Department memos suggested the delay contributed to the patient's death, subsequent investigations clarified that the patient's fatal injuries were caused by a city bus and that the autonomous vehicles were not responsible for the fatality. The incident highlights ongoing tensions and safety concerns regarding the integration of autonomous vehicles into urban emergency response environments.
In an initial report, a Cruise robotaxi was said to have delayed a San Francisco ambulance transporting Sammy Davis, a critically injured 69-year-old hit by a city bus. Davis later died. Subsequent clarification revealed that Cruise was not at fault for the fatality; the actual cause was a human-operated city bus. Despite this, the incident is included as it highlights challenges in the interaction between autonomous vehicles and emergency services in urban settings.
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 reports highlight regulatory gaps, noting that companies were only required to report collisions to regulators, not other interference incidents, reflecting inadequate oversight frameworks.
The Cruise autonomous vehicles failed to properly navigate or yield to emergency responders, demonstrating a lack of capability or robustness in complex real-world urban environments.
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 blockage was caused by the autonomous decision-making of the deployed Cruise vehicles, which unintentionally stopped and obstructed traffic post-deployment.
Added categories peak at 2: Minor (Epistemic Harm, Indirect)
Two Cruise autonomous vehicles temporarily blocked an ambulance in San Francisco, highlighting safety and integration challenges for autonomous systems in urban environments. The incident represents a minor, localized physical security concern with negligible national security implications, as it was unintentional and did not affect critical national infrastructure.
The Cruise vehicles operate independently in real-time traffic, responding to environmental inputs using learned patterns without explicit human planning or direct control. The report describes them as 'driverless taxis' and 'autonomous vehicles' that stopped and loitered in the street.
High Risk: The AI system is deployed in critical infrastructure (traffic management and road transport). The report describes 'autonomous vehicles' operating as 'driverless taxi services' which interfered with emergency responders.
The primary function of the Cruise vehicles is autonomous driving to transport passengers in urban environments.
The reports provide detailed, contrasting perspectives from both the San Francisco Fire Department and Cruise, including subsequent clarifications from city officials that resolved the initial ambiguity regarding the cause of the patient's death.
An alternative interpretation is that the Cruise vehicles did not cause any material delay, as Cruise claimed other vehicles were able to pass and the ambulance left 90 seconds after loading.
Technology
AV Inability to Yield
Two autonomous vehicles stopped in lanes and failed to clear ambulance path.
Reliance on Remote Assistance
Vehicles required manual remote intervention to navigate away from the scene.
Data Inputs
Inadequate Scene Context
AVs did not recognize the active emergency scene and blockages around them.
Human Factors
Delayed Remote Assistance
Remote operator took time to interface with police and move the vehicles.
Process and Methods
Deficient Emergency Protocols
Lack of effective procedures for AVs to yield to emergency responders.
Management
Rapid Deployment Decisions
Management pushed for fleet expansion before resolving emergency vehicle issues.
Regulatory Environment
Premature Service Expansion
State regulators approved 24/7 operations despite city safety protests.
Limited Incident Reporting
Regulators only required reporting of collisions, omitting blocking events.
AI system involved
Identified by automated extraction; can differ from the alleged developers above.
