Tesla Reports Record Monthly Driver-Assist Crash Total
Tesla reported 207 crashes involving Autopilot or Full Self-Driving-related Level 2 driver-assist use in May 2026, according to AutoGuide. The outlet reports that this was Tesla's highest monthly total in the dataset and exceeded its reported 157 crashes for all of 2021. Public records lack exposure data needed to calculate a per-mile crash rate.
Tesla reported 207 crashes involving its Level 2 driver-assistance systems in May 2026, the highest monthly total in the company's crash reports submitted to the National Highway Traffic Safety Administration, according to AutoGuide. The reported total covers incidents associated with Tesla's Autopilot and Full Self-Driving branding, although the public data described by AutoGuide does not allow a clean separation of crashes by software feature.
NHTSA's Standing General Order requires automakers to report certain crashes when an advanced driver-assistance system was engaged within roughly 30 seconds of the impact. AutoGuide reports that the reporting threshold includes incidents involving an airbag deployment, a tow-away, injury, fatality, or a pedestrian or cyclist strike.
What the reported data shows
AutoGuide reports that Tesla had 3,763 unique driver-assist crash reports, representing about 85% of industry ADAS reports in the dataset. It also reports that Tesla recorded 1,043 such crashes during 2025 and 826 through the first six months of 2026.
The May figure is larger than Tesla's reported 157 crashes across all of 2021, according to AutoGuide. Counts alone, however, do not establish whether crash risk per driven mile has risen or fallen. More vehicles on the road, more FSD subscriptions, and more engaged miles can all change the total number of reported crashes.
Exposure and transparency limits
AutoGuide reports that Tesla does not publish verified, independently auditable mileage data for Autopilot and Full Self-Driving use. Without feature-specific exposure data, external analysts cannot calculate a comparable crash rate per assisted mile from the public reporting file.
The outlet also reports that Tesla has redacted crash narratives on 99.9% of its 3,763 reports in each year since 2019, citing confidential business information, and that software-version fields are redacted. Those omissions limit independent analysis of individual incidents and whether they involved Autopilot or Full Self-Driving.
The episode illustrates a recurring measurement problem in deployed autonomy and driver-assistance systems: event totals are useful for identifying reporting volume and possible changes over time, but they are not a substitute for exposure-normalized safety metrics. Comparable evaluation generally requires clear definitions of feature engagement, operational design domain, miles or trips, and the severity distribution of outcomes.
Key Points
- 1Tesla's reported 207 May crashes set a monthly high in NHTSA's driver-assist reporting dataset, raising questions about the underlying safety evidence.
- 2Raw crash counts cannot establish per-mile risk because publicly available records do not include independently auditable Autopilot or Full Self-Driving exposure data.
- 3Safety evaluation across automated systems generally requires feature-specific exposure, operational context, and severity-normalized outcomes rather than incident totals alone.
Scoring Rationale
The reported record crash count is relevant to practitioners building, evaluating, and governing driver-assistance systems, particularly because it exposes limits in publicly available safety measurement. Its impact is moderated by the absence of exposure-normalized data and by the source report being more than three days old.
Sources
Primary source and supporting public references used for this report.
Practice with real Ad Tech data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Ad Tech problems
