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

Tesla in self-driving mode struck a parked Fullerton police cruiser blocking a fatal-crash scene while the driver looked at a phone

AIR-2024-0019 AIR record · 2024-06-tesla-autopilot-fullerton-police-vehicle
A real party was affected. This incident caused actual consequences, not a demonstrated capability.
From the supporting source
A Tesla was in its self-driving mode when it crashed into a parked patrol vehicle responding to a fatal crash in Orange County Thursday morning, police said.ktla.com
The record
Disclosed
2024-06-13
Occurred
2024-06-13
Vendor
Tesla
Product
Tesla Autopilot / Full Self-Driving
Risk domain
Agent behaviour › Unsafe actuation
System type
Classical autonomy (no generative model)
Who acted
Agent acted — the driving system struck the parked police cruiser
Attack surface
Autonomous operations
Autonomy level
Semi-autonomous
Initial vector
None autonomous action
Impact type
Physical safety, Property damage
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 police vehicle destroyed; officer narrowly escaped injury at an active fatal-collision scene
Capability abused
autonomous_drivingperceptionphysical_actuation
Defense & remediation
Recorded defense
driver-monitoring nag
Guardrail kinds
Human supervision
Guardrail outcome
Silent
Fix shipped
none attributed
Taxonomy & references
no_adversaryautomation_complacencystationary_obstacleemergency_services_interferencephysical_world
Curator notes
Two failure modes stacked: the system did not detect a lit, stationary emergency vehicle, and the driver had transferred attention to a phone because he relied on it. The second is the one that generalises to every supervised agent — supervision decays exactly as the agent gets good enough to trust. AI Incident Database incident 707.
Cite as
Enkrypt AI. "Tesla in self-driving mode struck a parked Fullerton police cruiser blocking a fatal-crash scene while the driver looked at a phone". AIR, AIR-2024-0019. https://agentincident.vercel.app/i/AIR-2024-0019/ (accessed 2026-10-19).