Uber autonomous vehicle kills pedestrian in Tempe Arizona

An Uber autonomous test vehicle struck and killed a pedestrian in Arizona due to a combination of AI classification errors and human operator inattention. The NTSB cited a lack of safety redundancies, including the deactivation of factory emergency braking systems.

Uber · Incident Mar 18, 2018 · Indexed Jun 9, 2026 · 2 sources

The system's inability to consistently classify the pedestrian prevented it from predicting her path and braking in time.
What
An Uber autonomous test vehicle struck and killed a pedestrian in Arizona due to a combination of AI classification errors and human operator inattention.
Incident date
Mar 18, 2018
Who
Uber
Failure mode
Agentic Action Error
AI surface
Autonomous System
Severity
Catastrophic

What happened

An Uber self-driving test vehicle struck and killed Elaine Herzberg as she walked across a road in Tempe, Arizona. The vehicle's operator was visually distracted by a cell phone and failed to intervene. The incident occurred at night in an area without a crosswalk.

What broke inside the model

Failure path · this incident · Agentic Action Error
  1. 01 · TriggerA pedestrian walks a bicycle across a dark road outside a crosswalk.
  2. 02 · Model stepThe perception system cycles her classification: vehicle, unknown object, bicycle, never settling.
  3. 03 · Control gapEach reclassification resets path prediction, and the factory emergency braking is disabled while the safety driver looks away.
  4. 04 · FailureBraking begins 1.2 seconds before impact.
  5. 05 · ConsequenceElaine Herzberg is killed; the first pedestrian death by a production autonomous vehicle.

The AI system failed to consistently classify the pedestrian, alternating between labeling her as a vehicle, an unknown object, and a bicyclist. Because the system did not maintain a consistent classification, it failed to predict the pedestrian's path and did not initiate braking until 1.2 seconds before impact.

Public visibilityHigh
Regulatory exposureNone
Customer impactClass-wide
Financial impactUnknown
Time to disclosureHours
  1. PrimaryAccident Report - NTSB/HAR-19/03ntsb.gov
  2. PressDeath of Elaine Herzberg - Wikipediaen.wikipedia.org
Permalinkhttps://failureindex.ai/failures/uber-autonomous-vehicle-kills-pedestrian-tempe
CitationAI Failure Index. "Uber autonomous vehicle kills pedestrian in Tempe Arizona" (FI-0329). Realm Labs. https://failureindex.ai/failures/uber-autonomous-vehicle-kills-pedestrian-tempe (indexed Jun 9, 2026).
Share cardA branded image of this record for posts and slides.

Data fields CC-BY 4.0, prose citation permitted. Incident ID FI-0329. Full dataset at /data.

Note from Realm Labs, the Index steward

How Realm fits

Controls for this failure mode
  • Prism
  • OmniGuard
  • AgentRealm

This entry sits in the index's predictive wing: a system that scores, ranks, perceives, or steers rather than generates. Realm's runtime layer is built for the generative and agentic systems now moving into these same decision seats, where it watches a model's internal state and holds an unsupported claim or an unchecked action before it commits. The control gap on this record, an automated decision that reached people with no runtime check in front of it, is the same gap. The index keeps predictive failures on the record because the pattern carries straight into the systems shipping today.