Agent Incident RegistryAIR
v1 · updated 19 Oct 2026
Safety failure Realized harm

A Waymo struck a child running from behind a double-parked SUV toward an elementary school during drop-off, prompting an NHTSA probe

AIR-2026-0023 AIR record · 2026-01-waymo-santa-monica-child-struck
A real party was affected. This incident caused actual consequences, not a demonstrated capability.
From the supporting source
The agency said there were other children, a crossing guard, and several double-parked vehicles in the vicinity.aljazeera.com
The record
Disclosed
2026-01-29
Occurred
2026-01-23
Vendor
Waymo (Alphabet)
Product
Waymo 5th Generation Automated Driving System
Risk domain
Agent behaviour › Unsafe actuation
System type
Classical autonomy (no generative model)
Who acted
Agent acted — the vehicle struck the child during school drop-off
Attack surface
Autonomous operations
Autonomy level
Fully autonomous
Initial vector
None autonomous action
Impact type
Physical safety
Confidence
High
Submitted by
EAI Internal
Crosswalk
OWASP Agentic Top 10 (2026)
ASI01 – Agent Goal Hijack
MITRE ATLAS
AML.T0048 – External Harms
NIST AI RMF
MEASURE 2.6; MANAGE 1.3
EU AI Act
Annex III §1(c)
ISO/IEC
TR 24028; 24029-1; 23894
AIUC-1
C. Safety
Scope
agent misuse
Mitigation
Require confirmation for destructive actions; Implement dry-run modes; Use layered safety controls.
Taxonomy
Enkrypt AI agent risk taxonomy · Black-Box Red Teaming of Agentic AI: A Taxonomy-Driven Framework for Automated Risk Discovery
Blast radius
one child with minor injuries; NHTSA investigation PE26001 into the 5th-generation ADS
Capability abused
autonomous_drivingocclusion_reasoningcontext_aware_speed
Defense & remediation
Recorded defense
pedestrian detection and school-zone behaviour
Guardrail kinds
Physical safety system
Guardrail outcome
Silent
Fix shipped
none at disclosure; federal investigation opened
Taxonomy & references
no_adversaryocclusionchildren_at_riskcontext_aware_cautionregulatory_investigationphysical_world
Curator notes
Every cue for extreme caution was present and legible — a crossing guard, other children, double-parked cars creating occlusions, school drop-off hours — and the agent's speed and margin did not reflect them. Occlusion is unavoidable; failing to infer higher risk from an obviously child-dense scene is the defect. Self-reported by Waymo to NHTSA. AI Incident Database incident 1361.
Cite as
Enkrypt AI. "A Waymo struck a child running from behind a double-parked SUV toward an elementary school during drop-off, prompting an NHTSA probe". AIR, AIR-2026-0023. https://agentincident.vercel.app/i/AIR-2026-0023/ (accessed 2026-10-19).