Google Waymo Vehicle and Bus Allegedly Collide in California

On February 14, 2016, a Google autonomous test vehicle partially responsible for a low-speed collision with a bus on El Camino Real in Google’s hometown of Mountain View, CA.

Risk classification

• Primary risk domain: 7 AI system safety, failures, & limitations
• Primary risk subdomain: 7.3 Lack of capability or robustness

The AI system failed to accurately predict the behavior of the public bus, demonstrating a lack of capability and robustness in handling complex real-world driving negotiations.

Additional risk subdomains

• 5.1 Overreliance and unsafe use: The human test driver failed to intervene despite seeing the bus, relying on the assumption that the AI system's prediction was correct.

Causal factors

• Entity: AI
• Intent: Unintentional
• Timing: Post-deployment

The collision was caused by an incorrect prediction made by the AI system's software during autonomous operation on public roads post-deployment.

EU AI Act risk tier

• Risk tier: 2 High Risk

High Risk: The AI system is used in critical infrastructure (traffic management and autonomous road transport), which has significant implications for public safety.

AI system and alleged parties

• AI system: Google self-driving car (Google)
• AI purpose: Autonomous Driving; Navigation Assistant
• Behaviour type: Autonomous
• Alleged developer: Google
• Alleged deployer: Google
• Alleged harmed parties: Mountain View municipal bus passengers, Mountain View municipal bus

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Negligible. Severity is scored from Negligible to Catastrophic in each harm category, for harm the reports describe as caused directly or indirectly by the AI system.

• Physical: direct Negligible, indirect Negligible
• Infrastructure: direct Negligible, indirect Negligible
• Property: direct Negligible, indirect Negligible
• Financial: direct Negligible, indirect Negligible
• Environmental: direct Negligible, indirect Negligible
• Malicious content: direct Negligible, indirect Negligible
• Differential treatment: direct Negligible, indirect Negligible
• Civil rights: direct Negligible, indirect Negligible
• Democracy: direct Negligible, indirect Negligible
• Privacy: direct Negligible, indirect Negligible
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Negligible, indirect Negligible
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

People affected

• Occurrences reported: 1
• People reportedly exposed: 17

Potential causes

Management

• Aggressive Deployment Goals: Pushing for fully driverless vehicles before solving social driving norms.
• Inadequate Risk Mitigation: Management failed to prevent unsafe assumptions in the lane-merging logic.

Technology

• Incorrect Yield Prediction: Software predicted the bus would yield because the car was ahead of it.
• Flawed Lane Merging Algorithm: Algorithm failed to safely merge back into traffic after dodging sandbags.
• Software Tweak Side Effects: Recent update to hug the right curb created complex lane-merging scenarios.

Data Inputs

• Obstacle Detection (Sandbags): Sandbags around a storm drain forced the vehicle to deviate from its lane.
• Size-Based Yield Assumptions: System lacked deep understanding that large vehicles are less likely to yield.
• Unexpected Road Obstacles: Sandbags around storm drain required sudden lane-merging maneuver.

Human Factors

• Safety Driver Inaction: Test driver assumed the bus would slow down and did not take manual control.
• Overreliance on Automation: Driver trusted the vehicle's decision-making instead of intervening early.
• Bus Driver Expectation: Bus driver assumed the Lexus would stay stationary and did not yield.

Process and Methods

• Inadequate Social Motoring Rules: System lacked logic for the informal give-and-take negotiation of driving.
• Insufficient Edge Case Testing: Testing did not fully cover interactions with large municipal buses.
• Public Road Software Testing: Testing unproven software tweaks directly on busy public roads.

Regulatory Environment

• Lack of Autonomous Standards: Absence of clear federal performance standards for self-driving decisions.
• Permissive Testing Rules: California DMV allowed testing of autonomous modes with soft oversight.

Information quality

• Classification confidence: High
• Reason for confidence: The reports provide consistent, detailed accounts of the collision on February 14, 2016, including official DMV accident reports, statements from Google, and details from the transit authority. There is high consensus on the facts of the primary incident.
• Ambiguities identified: Some reports in the text bundle describe a completely different accident on September 23, 2016, where a van ran a red light, which could cause confusion if not carefully separated from the February 14, 2016 bus collision.
• Alternative interpretations: None for the primary incident, as Google explicitly acknowledged partial responsibility for the bus collision.

A Google self-driving Lexus SUV collided with a public bus at low speed in Mountain View, California, marking the first time an autonomous vehicle was found at fault. The incident resulted in minor property damage, no injuries, and carries negligible national security implications across all assessed categories.

• Overall national security impact: Negligible
• Scope: Single nation
• Primary target: No clear primary
• Alleged perpetrator: Unknown

Threat characteristics

• Imminence: Long-term. The incident represents a localized, post-deployment software failure rather than an active national security crisis.
• Autonomy: Human-supervised. The vehicle operated autonomously but with a human test driver on board who had the capability to intervene but failed to do so.
• Novelty: Evolved capability. This marked the first time an autonomous vehicle was found at fault in a collision, representing an evolution in autonomous system failure modes.

Impact by dimension

• Physical security: Negligible. The incident was a minor traffic collision involving a test vehicle and a public bus with no injuries or critical infrastructure disruption, posing negligible threat to national physical security.
• Information security: Negligible. No intelligence compromise, classified data theft, or information warfare operations were associated with this localized traffic accident.
• Sovereignty: Negligible. The accident had no impact on state authority, electoral systems, or core government decision-making processes.
• Economic security: Negligible. Financial impact was limited to minor property damage of 2,000 to 3,000 dollars for bus repairs, representing zero threat to strategic economic or technological security.
• Societal stability: Negligible. The incident was a routine traffic accident with no implications for civil liberties, societal stability, or human rights.