---

---

As we’ve reported, a trial has begun in Florida related to a crash that happened here a handful of years ago while the Tesla was on Autopilot. The driver of the car did not die in the crash, but a bystander was. “Attorneys for the estate of Benavides Leon contend that the Tesla Autopilot system malfunctioned and that malfunction was a proximate cause of the crash,” Steve Hanley wrote in a long and detailed explanation of the case.

The trial started on June 14, and, I have to say, CleanTechnica has not been in the courthouse listening to the arguments on both sides, and I wasn’t even sure if we’d end up covering this again, but I have received some notes and highlights regarding the opening days of the trial, and they really pulled me in and come with some interesting insights, so I figured others would like to see these notes as well. Here’s what RebuttalPR sent over:

---

TRANSCRIPTS: TESLA TRIAL OVER AUTOPILOT CRASH; DAYS 1, 2 AND 3 COMPLETE

The first third-party wrongful death trial against Tesla began in Miami’s federal courthouse on Monday, July 14th.

As the first three days wrapped up, we heard opening arguments from both sides and testimonies for the Emergency Medical Services responder and Florida Highway Patrol Officer on-scene as well as Dr. Mendel Singer, an expert in biodata statistics and analysis. The jury also heard parts of a deposition of Tesla Autopilot Firmware Engineer Akshay Phatak. On Wednesday, former NHTSA Senior Safety Advisor Mary Louise “Missy” Cummings took the stand for a full day of testimony and cross examination.

Here are some of the key moments:

Trial transcripts for the first three Days of the trial are available at this Dropbox link.

Corporal David Riso, lead traffic homicide investigator with the Florida Highway Patrol

During his investigation, Riso tried to retrieve the autonomous data from Tesla but never received it. Riso had taken the computer components from the vehicle after coordinating with the NTSB.

• When trying to retrieve the data, the Tesla technician told Riso that the file was corrupted. (Day 2 AM, 57)
• He also said that the Tesla technician’s sworn affidavit about the data retrieval process was “untrue” (Day 2 AM, 59)
• Had he been given the data, Riso said “I would have incorporated it in my report. I also would have contacted the NTSB and US DOT who were inquiring about the crash, I would have talked to my supervisor first and let him know to the give the data to the government agency that wanted it.” (Day 2 AM, 60)

Dr. Mendel Singer, expert statistician and Professor at Case Western University School of Medicine

When asked to assess the validity of Tesla’s Vehicle Safety Report:

• “I am not aware of any published study, any reports that are done independently…where [Tesla] actually had raw data and could validate it to see does it tend to make sense.” (Day2 PM, 79)
• He elaborated on the discrepancies, noting that the “vast majority for the crashes for non-Teslas that they are counting or crashes that they wouldn’t count in a Tesla, because non-Teslas crashes are counted based on police reports, regardless of safety system deployment. (Day 2 PM, 81)
• “[Tesla] incorrectly stated on the website that they’re only counting crashes with light vehicles under 10,000 pounds, but basically more passenger vehicles, but that is actually not true. If you actually do the math from the reports they actually use the miles and crashes for all vehicles another road whether it’s a bus, whether it’s a semi truck whether it’s a mistake it’s just all lumped together.” (Day 2 PM, 83)

Discussing Tesla’s crash data from their website:

• “There were some mistakes. So on the website they just – theydon’t have the old numbers on the website. They just changed all the numbers going back to update it so that these corrections are made in all the data they’re saying is now correct now, but was wrong for the last four years.” (Day 2 PM, 85)
• “There’s no explanation, you know, why was the data wrong for four years? And there’s no explanation given why the data was so wrong for Autopilot off but not for Autopilot on.” (Day 2 PM, 86)
• “The question gets back to validating data. So if the total miles are wrong, why didn’t they validate it and check? Let’s add up the miles Autopilot off and on, add it up does that equal the total miles. Why wouldn’t they have checked that? Why is it that it can be wrong, you know, so wrong.” (Day 2 PM, 87)
• Attorneys for the plaintiffs asked Singer: “So is it fair to say if Tesla operating opposing Autopilot blasts threw an intersection, hitting a toddler, who then goes flyer, but pre-tensioner doesn’t fire or air bag doesn’t deploy, under Tesla’s crash counts, that’s not a crash?
  • Singer replied: “That’s correct.” (Day 2 PM, 80)
• Singer unequivocally stated that if Tesla’s report were submitted to a peer-reviewed journal he edited, “That would have been a really quick and easy rejection.” (Day 2 PM, 90)

Deposition of Akshay Phatak, Autopilot Firmware Engineer at Tesla Motors (Day 2 PM)

• Phatak was not in the courtroom, so his deposition was played.
• Phatak was asked if prior to 2018 Tesla maintained any records on safety levels of their vehicles with and without Autopilot engaged:
  • “We did not have a specific documentation for crashes, yes.” (Day 2 PM, 140)
• Prior to March of 2018, did Tesla maintain any records or data that kept track of the number of crashes that occurred per vehicle mile driven with Autopilot engaged?
  • “No.” (Day 2 PM, 141)

• On day 3, Dr. Missy Cummings was asked if there was any Tesla data supporting their claims that they had the most robust set of warnings against driver misuse or abuse ever deployed. She was asked if it was feasible for them to improve their warnings in the years prior to this case’s crash.
  • “Correct. In Phatak’s deposition, it was clear they knew that they had a big problem with people ignoring the warnings. Ignoring the hands-on requests. And…as you know, prior to this accident. It was known to Tesla that they were having problems with people ignoring their warnings.” (Day 3 AM, 61)

Mary Louise “Missy” Cummings, former National Highway Traffic Safety Administration Senior Safety Advisor (Day 3 AM/PM)

• “It is my professional opinion that Tesla’s Autopilot is defective because Tesla knowing allows the car to be operated in operational domains for which it is explicitly not designed for.” (Day 3 AM, 28)
• When asked if she has “any opinion as to why Tesla chose not to geofence its technology in 2019”, Missy Cummings said “I believe that they were using that as a way to sell more cars.” (Day 3 AM, 52)
• After a deposition of Tesla Corporate Representative Eloy Rubio Blanco was played where Rubio Blanco admitted that “as of 2019 Automatic Emergency Braking would not fire in the scenario of a T intersection collision,” Missy Cummings said the limitations of AEB were not disclosed to Tesla vehicle owners (Day 3 PM, 35)
• When asked if what Elon Musk said in 2016 that a Tesla production vehicle could “drive safer than a human,” Cummings said “it wasn’t true then and it isn’t true now.” (Day 3 PM, 71-72)

Proceedings in Benavides v. Tesla began on Monday, July 14, 2025 at the Wilkie D. Ferguson, Jr. U.S. Courthouse in front of the Honorable Judge Beth Bloom. Dillon Angulo and the family of Naibel Benavides are represented by lead attorney Brett Schreiber alongside attorneys Adam Boumel and Todd Poses.

---

There are a number of interesting, or you could say odd, things in there about Tesla’s Autopilot crash data, and the lack of 3rd party data access or analysis.

It’s not clear if this is really just a one-off case or if there are going to be some important implications for Tesla and its driver-assist systems. We’ll see….