Cruise autonomous vehicles repeatedly drove toward or over fire hoses during active firefighting operations in San Francisco, requiring firefighters to physically intervene including breaking vehicle windows to prevent interference with emergency operations.
Between April 2022 and January 2023, Cruise autonomous vehicles operated by General Motors subsidiary Cruise LLC repeatedly interfered with San Francisco Fire Department operations. The most serious incident occurred on January 21, 2023, when a Cruise AV entered an active fire scene on Hayes Street near Divisadero Street and approached fire hoses on the ground. Firefighters had to shatter the vehicle's front window to prevent it from driving over the hoses, which could seriously injure firefighters and violates California Vehicle Code Section 21708. Similar incidents occurred in June 2022 when a Cruise AV ran over a fire hose in use, and in April 2022 when a vehicle stopped in a travel lane obstructing a fire truck responding to a 3-alarm fire. The San Francisco Fire Department reported at least 39 interference incidents in 2023, including robotaxis blocking firehouses, driving through police tape, and stopping in active emergency scenes. Only Cruise experts can disengage vehicles from autonomous mode, requiring firefighters to contact remote operators or wait for human employees to arrive. These incidents prompted San Francisco transportation officials to request that regulators halt approval for expanded Cruise operations until safety issues are resolved.
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.
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