A Note on Tesla's Revised Safety Report Crash Rates
URL: http://arxiv.org/abs/2311.06187v2
archive: archived pipeline: cataloged verified
Abstract
Between June 2018 and December 2022, Tesla released quarterly safety reports citing average miles between crashes for Tesla vehicles. Prior to March 2021, crash rates were categorized as 1) with their SAE Level 2 automated driving system Autopilot engaged, 2) without Autopilot but with active safety features such as automatic emergency braking, and 3) without Autopilot and without active safety features. In January 2022, Tesla revised past reports to reflect their new categories of with and without Autopilot engaged, in addition to making small adjustments based on recently discovered double counting of reports and excluding previously recorded crashes that did not meet their thresholds of airbag or active safety restraint activation. The revisions are heavily biased towards no-active-safety-features$\unicode{x2014}$a surprising result given prior research showing that drivers predominantly keep most active safety features enabled. As Tesla's safety reports represent the only national source of Level 2 advanced driver assistance system crash rates, clarification of their methods is essential for researchers and regulators. This note describes the changes and considers possible explanations for the discrepancies.
Summary
Short communication analyzing Tesla's January 2023 revision of its quarterly Vehicle Safety Reports. Tesla replaced its three-category reporting (Autopilot engaged; without Autopilot but with active safety features; without Autopilot and without active safety features) with a two-category system (using Autopilot; not using Autopilot) and retroactively changed crash rates back to Q3 2018, citing discovery of duplicate reports, events without airbag/restraint deployment, and invalid mileage records. The revisions modestly reduced reported Autopilot crashes per 100M miles (average -3.6% across Q3 2018-Q4 2021) but produced large, asymmetric shifts in non-Autopilot rates, with the new 'not using Autopilot' category aligning closely with the prior 'no active safety features' subgroup despite evidence that the vast majority of drivers leave active safety features (e.g., automatic emergency braking, default-on each ignition) enabled. Goodall argues the revised non-AP rates therefore appear inflated relative to a representative Tesla fleet, biasing comparisons of Autopilot vs non-Autopilot safety. Because Tesla's reports are the only national source of Level 2 ADAS crash-rate data, the note calls for greater methodological transparency.
Key finding
Tesla's January 2023 retroactive revision left Autopilot crash rates roughly unchanged (mean -3.6%) but shifted the non-Autopilot category to closely match the prior 'no active safety features' subgroup, despite HLDI evidence that ~93% of drivers keep front crash prevention enabled and that Tesla's automatic emergency braking is on by default each ignition; the revised non-AP benchmark therefore appears biased upward, distorting Autopilot-vs-non-AP safety comparisons.
Methodology
Archival/observational analysis of publicly reported crash-rate statistics. Pre- and post-revision quarterly Tesla Vehicle Safety Report figures (Q3 2018 through Q4 2021) were retrieved via the Internet Archive Wayback Machine and tabulated to compute per-quarter percentage changes in crashes per 100 million miles for Autopilot and non-Autopilot categories. Findings are interpreted against external evidence on driver use of active safety features (HLDI service-department audit) and Tesla owner's manual specifications for default activation of automatic emergency braking.
Quality score: 5 / 5