236 crashes, four fatalities: Tesla blacks out key evidence citing ‘financial harm’

Reports of crashes involving Tesla’s assisted-driving systems surged to 236 in July 2026, the highest monthly figure recorded by the US National Highway Traffic Safety Administration (NHTSA) since it introduced mandatory reporting in 2021. Four were fatal incidents, occurring in Maryland, California and Minnesota, and are known to have caused at least seven deaths. More notably, Tesla blacked out key fields in the reports it submitted to regulators, including the software version, descriptions of what happened and whether the vehicle was operating within the system’s recognised operating parameters, citing the possibility of “serious financial harm” to the company.

The figures raise questions on two fronts. First, as autonomous driving technology becomes increasingly widespread, are commercial considerations eroding consumers’ right to know? Second, can the existing data-disclosure framework genuinely constrain the behaviour of carmakers?

Crash figures nearly double in four months

Data collected under the NHTSA’s Standing General Order shows a clear stepwise increase in crashes that occurred while Tesla’s Autopilot or Full Self-Driving (FSD) was engaged: 135 in March, rising to 207 in May, edging up to 209 in June and reaching a new high of 236 in July.

An even more striking comparison is that Tesla’s assisted-driving systems were linked to just 157 crashes in the whole of 2021. The figure for a single month has now surpassed the total for that entire year.

The locations and casualties in the four fatal incidents reported in July were:

Telemetry data confirmed that an assisted-driving function was engaged at the time of each incident. However, Tesla’s submitted reports place significant barriers around key fields that could help trace the details of what happened.

Transparency blocked by ‘financial harm’ claim

Under US federal rules, crash reports submitted to the NHTSA must include an account of the incident, the software version number and whether the crash occurred within the system’s approved design operating domain. Tesla applied for an exemption on the grounds that releasing the information would cause the company “serious financial harm”. As a result, the full details of all four fatal cases were redacted. This is not the first time: over the past several years, Tesla has partially concealed similar fields in multiple reports submitted to the NHTSA, citing the same reason.

By contrast, crash data submitted for General Motors’ Super Cruise system did not receive similar treatment in the narrative field. Ford’s BlueCruise also showed no comparable level of redaction.

It is also worth noting that Tesla is currently the subject of a separate NHTSA investigation into how it reports Autopilot and FSD crashes. The redaction practice itself is one of the issues being examined by regulators — a key piece of background when assessing the seriousness of the matter that was not mentioned in the original text.

The statistics have blind spots too: death toll may be understated

To understand the figures properly, it is first necessary to clarify what they represent. They are not statistics covering all traffic accidents. Under the NHTSA order, carmakers must report a crash if an assisted-driving system was engaged at any point in the 30 seconds before the collision. Ordinary crashes in which assisted driving was not activated are excluded.

More importantly, each report records only one person with the “highest level of injury severity”. That means the Maryland crash, which actually caused three deaths, appears in the database as a single death. The Minnesota case involving two deaths is likewise counted as one incident. In other words, the true death toll may be higher than the seven deaths disclosed, and any statistics based on these figures — including safety comparisons cited by Tesla itself — start from an understated baseline.

Whether the figures reflect an increase in the number of vehicles involved in crashes or a rise in the road mileage driven with assisted driving engaged requires mileage data. Such data is not publicly available.

Gap between marketing and legal responsibility

Tesla’s consumer-facing marketing has consistently carried strong suggestions of “full self-driving”. As a result, some owners have developed a level of trust closer to that associated with autonomous driving and have been willing to reduce their active monitoring of the vehicle for extended periods. Yet Tesla has maintained one position in its legal disclaimers and in assigning responsibility for crashes: unless the system has formally obtained SAE Level 4 certification or above, the driver remains solely responsible.

This creates a mismatch: the product continues to raise users’ expectations, while the regulatory system has not established a corresponding accountability mechanism.

Who bears the consequences?

In the short term, Tesla owners are likely to bear the brunt, particularly those who have bought the FSD package. The one-off purchase price for FSD is currently US$8,000, or US$99 a month by subscription (the original text described it as about US$12,000; the current price has since been lowered, as noted in the fact-check below). As negative news accumulates, insurers may reassess premium models for Tesla vehicles and add costs associated with assisted driving.

For institutions investing in Tesla and the wider autonomous-driving supply chain, the figures offer an important window into whether there is an increasingly difficult-to-reverse gap between technological maturity and the speed of commercialisation. If regulators eventually move from voluntary disclosure to mandatory, quantifiable safety standards, it remains unknown whether Tesla can absorb the higher costs of redundant safety design in the same way as traditional carmakers.

Three indicators to watch