Skip to main content
Incident intelligence/SS-IR-029CASE FILE OPEN
Symbolic editorial illustration for SS-IR-029SERVANTSTACK // INCIDENT INTELLIGENCEFORENSIC IMAGE // VERIFIED FRAME
SS-IR-029 // INCIDENT REPORTDocumented

Tesla

DOJ Opens Criminal Probe Into Self-Driving Marketing Amid Autopilot Deaths

EXECUTIVE BRIEF

On October 26, 2022, Reuters reported that the U.S.

AUTONOMOUS & AVIATION
FAILURE CHAINTRACE COMPLETE
  1. 01TRIGGEROn October 26, 2022, Reuters reported that the U.S.
  2. 02MACHINE ACTIONAutonomous actor
  3. 03MISSING GATEExecution gate and human override
  4. 04IMPACTPhysical safety
01 // INCIDENT SUMMARY

The short version

On October 26, 2022, Reuters reported that the U.S.

02 // KEY FACTS

Case telemetry

INCIDENT
SS-IR-029
DATE
October 26, 2022
SYSTEM
Tesla
LOCATION / SCOPE
United States (DOJ offices in Washington, D.C. and San Francisco)
EVIDENCE
Documented
AI ROLE
Autonomous actor
HARM
Physical safety
SOURCES
3 cited records
03ENTRY POINT // WHAT HAPPENED

The event

On October 26, 2022, Reuters reported that the U.S. Department of Justice had opened a criminal investigation into Tesla, with prosecutors in Washington, D.C. and San Francisco examining whether the company misled consumers and investors by marketing its Autopilot and Full Self-Driving (FSD) software as capable of driving the car itself. The probe followed more than a dozen crashes in which Autopilot was active, several of them fatal. Tesla's marketing went back years: a 2016 promotional video on the company's site declared, "The person in the driver's seat is only there for legal reasons. He is not doing anything. The car is driving itself," and Elon Musk publicly called the system "probably better" than a human driver. The probe landed amid a wave of Autopilot fatalities, including two motorcyclist deaths in mid-2022: Landon Embry, 34, killed when a Tesla Model 3 with Autopilot confirmed engaged struck his Harley-Davidson from behind on Interstate 15 in Draper, Utah on July 24, 2022, and an unidentified rider killed when a Tesla Model Y hit a Yamaha on State Route 91 near Riverside, California on July 7, 2022. As of 2016, NHTSA had opened investigations into 39 crashes suspected of involving automated driver-assist systems; 30 of those involved Teslas, and they accounted for 19 deaths. NHTSA's separate parked-emergency-vehicle probe had by June 2022 been expanded to cover nearly every Tesla sold in the U.S. since 2014.

04CAUSAL TRACE // AI'S ACTUAL ROLE

What the machine did

Autopilot and FSD are SAE Level 2 driver-assist systems: the human is legally required to stay fully attentive with hands on the wheel at all times. Yet Tesla branded and marketed them with names and videos implying full autonomy, with no validation gate forcing the consumer-facing claims to match the engineering reality stated in the fine print. The result was a system that performed lane-keeping and adaptive control at machine speed while users, primed by the marketing, over-trusted it and disengaged from supervision. There was no human SME sign-off reconciling what the product was sold as against what it was actually certified and warned to do, so the gap between "the car is driving itself" and "intended for use with a fully attentive driver" went unchallenged into showrooms, ad copy, and ultimately fatal real-world driving.

Autonomous actorAutomation was a causal participant—not a decorative label for the system around it.
05BLAST RADIUS // CONSEQUENCES

Where the failure landed

A federal criminal probe by DOJ prosecutors in two offices, with possible outcomes ranging from criminal charges to civil sanctions to no action. Tesla shares fell on the report. The investigation compounded an existing NHTSA defect probe covering nearly all U.S. Teslas since 2014 and ran alongside a fatality record that, by NHTSA's accounting, included 19 deaths across 30 Tesla crashes suspected of automated-system involvement. Most consequentially, real people died: at minimum a confirmed Autopilot-engaged collision killed motorcyclist Landon Embry, and additional Autopilot-active fatal crashes were the predicate for the criminal inquiry.

06 // EVIDENCE STATUS

Documented

Supported by a first-party disclosure, technical research, or corroborated reporting cited below.

SOURCE RECORD UPDATED 2026-07-09

07 // SOURCE LEDGER

3 cited records

  1. 01
  2. 02
  3. 03
08CONTROL FAILURE // MISSING GOVERNANCE

Execution gate and human override

The failure pattern in this case: Autonomous high-consequence action.

09INTERVENTION POINT // HUMAN IN THE MIDDLE

The moment the path could change

A trained operator receives the evidence, owns the go/no-go decision, and retains an immediate override.

AI PROPOSES→HUMAN OWNS THE DECISION→SYSTEM EXECUTES
10CONTROL DEPLOYMENT // AUTHORITYGATE

High-consequence gate · human override

The AuthorityGate Operational Resilience framework requires a human SME marketing-and-claims validation gate: every customer-facing capability statement about an autonomous or AI-assisted system must be reconciled, before publication, against the system's certified operating envelope and safety documentation by a named domain expert with sign-off authority. A claim like "the car is driving itself" describing a Level 2 system whose own manual says "intended for use with a fully attentive driver" is exactly the contradiction the gate exists to catch -- the reviewer blocks any marketing asset whose autonomy claim exceeds the validated, warned operational design domain, and routes the mismatch back to engineering and legal for reconciliation rather than letting it ship. The gate also covers change-validation: any update that materially expands the asserted capability (a new "Full Self-Driving" label, an unsupervised demo video) triggers re-review before release, so the gap between what the product does and what the public is told it does can never widen unsupervised.

RELEVANT KEYSTONE CONTROLHuman-in-the-Loop ValidationHow high-risk actions route to a named subject-matter expert who owns the go or no-go decision.
12 // THE ALTERNATIVE

Autonomy is a design choice.

See the operating model that keeps AI useful while preserving human authority at consequential moments.

Compare AgenticAI and AugmentedAI →