This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
Zoox issued a voluntary software recall for 332 robotaxis after identifying that the autonomous driving system caused vehicles to cross into opposing lanes or stop in front of oncoming traffic near intersections. While no collisions were reported, the behavior was identified as a safety risk. The company has since deployed software updates to address the issue.
A Zoox robotaxi reportedly crossed partly into the opposing lane and stopped near oncoming traffic during a right turn on August 26, 2025. Zoox later recalled 332 automated-driving systems after identifying similar lane-crossing behavior; no collision was reported.
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.
The robotaxis exhibited a lack of capability and robustness by failing to maintain proper lane positioning and blocking crosswalks during autonomous operation.
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 risk was caused by the autonomous driving system making unexpected lane-crossing maneuvers while operating on public roads after deployment.
A voluntary software recall of 332 Zoox robotaxis due to lane-crossing anomalies presents negligible national security risk. The incident was unintentional, resulted in no collisions or injuries, and has been resolved via software updates under standard regulatory oversight.
The report describes robotaxis operating independently on public roads, responding to environmental inputs and making real-time driving decisions without human intervention.
High Risk: The autonomous driving system is deployed in critical transport infrastructure and safety-critical applications on public roads, posing significant implications for public safety.
The AI system's primary function is to autonomously navigate and drive robotaxis on public roads.
The reports provide clear, factual details about the software recall, the specific driving failures (crossing center lines, stopping in crosswalks), the number of affected vehicles (332), and the number of identified instances (62). There is no ambiguity about the lack of collisions or injuries.
None
Technology
Flawed Path Planning Software
Software caused wide right turns, crossing into oncoming lanes.
Suboptimal Intersection Logic
Vehicle stopped in crosswalks to avoid blocking intersections.
Poor Object Prediction Logic
Prior software versions struggled to predict other road users' movements.
Data Inputs
Inaccurate Lane Boundary Data
System misjudged center yellow lines and crosswalk boundaries.
Pedestrian Tracking Failures
Previous issues involved poor tracking of nearby pedestrian data.
Human Factors
Emulating Unsafe Human Habits
System performed wide turns common for humans but unsafe for AI.
Process and Methods
Insufficient Edge-Case Testing
Testing failed to prevent lane-crossing and crosswalk-blocking behaviors.
Inadequate Pre-Release Review
Software was deployed despite failing to meet internal safety standards.
Management
Delayed Risk Mitigation
First incident occurred in August but full recall was delayed to December.
Regulatory Environment
Regulator Pressure and Inquiry
NHTSA discussions and investigations pushed the voluntary recall.
AI system involved
Identified by automated extraction; can differ from the alleged developers above.
