Tesla didn’t just settle a wrongful-death lawsuit this week. It settled a case that was about to put a single word on trial — not a sensor, not a software build, not a line of code, but the word “Autopilot” itself. A company approaching a trillion-dollar valuation would rather write a confidential check than let twelve strangers spend a week debating whether its own product name is honest. That decision says more about Tesla’s legal exposure than the crash that started it all.
The crash happened on February 18, 2023, around 4 a.m., on Interstate 680 in Contra Costa County, California. Genesis Giovanni Mendoza Martinez was driving his 2014 Model S with Autopilot engaged when the car, still doing roughly 70 mph, plowed into a county ladder fire truck parked across the lanes at an earlier accident scene. Mendoza was killed. His brother Caleb, riding in the passenger seat, was seriously injured. It’s a grimly familiar shape of crash for anyone who has followed Autopilot litigation: a stationary vehicle in the roadway, and a car that never slowed down.
Related Articles
- This South Carolina ‘Widowmaker’ Tree Already Had a Body Count and a Nickname. Then It Killed Four More People In One Night.
- Sunroofs Are Shattering All Over America. Turns Out No One Ever Required Them Not To.
The Mendoza family’s case landed in the Northern District of California in front of U.S. District Judge Vince Chhabria, who has now handled years of Tesla’s Autopilot docket. This wasn’t a simple defect claim. The family argued that Tesla, and Musk personally, had misrepresented what Autopilot could actually do — and that the gap between the marketing and the machine is what put Mendoza behind the wheel with more confidence than the technology deserved.
In May 2025, Chhabria let that theory proceed toward trial. The order is blunt: the Autopilot name is “plausibly misleading,” and Musk’s own public statements about the system’s abilities were ruled fair game as evidence. Read that twice. A federal judge decided a jury could reasonably find that Tesla’s marketing department, not just its engineering department, helped create a hazard.
Here’s the detail that almost nobody outside aviation law thinks about: “autopilot” already had a precise, well-understood meaning before Tesla borrowed it in 2014. On an airliner, autopilot holds a heading, an altitude, or a flight path — and federal rules still require two type-rated pilots in the cockpit, actively monitoring it, ready to take over in seconds. Aviation autopilot has never meant nobody needs to pay attention. Tesla took a word whose entire professional meaning is constant supervision and marketed it to consumers as almost the opposite. That gap, between the old meaning and the new impression, is the actual subject of years of Autopilot litigation — not the radar, not the cameras, not the neural network underneath.
That’s the argument Tesla did not want in front of a jury, so it settled instead. This is now a pattern rather than a one-off. The Auto Wire has covered this kind of crash before, back when a driver died after his Tesla drove into a pond while Autopilot was steering. Tesla tends to resolve fatal Autopilot and Full Self-Driving cases quietly, before a jury ever reaches a verdict.
The one time this pattern broke, Tesla lost badly. A Florida federal jury returned a historic $243 million verdict in a 2019 Autopilot death case, and the judge refused to throw it out. Since then, Tesla has settled case after case rather than risk a repeat, including a fatal Full Self-Driving pedestrian crash resolved in June. Tesla is now said to be facing up to $14.5 billion in Autopilot and FSD-related lawsuits, and the plaintiffs’ bar, emboldened by the Florida verdict, shows no sign of slowing down.
Every confidential settlement does two things at once. It removes one case from the board, and it keeps the theory that “Autopilot” is misleading from ever becoming a published verdict other plaintiffs can point to. Tesla has used a similar playbook in disputes that have nothing to do with crashes. Earlier this year, it quietly settled a class action over Supercharger idle fees rather than let a court examine how those fees were calculated. Settling isn’t unusual for a large company. What’s notable is how consistently Tesla reaches for it the moment a case is about to test the accuracy of its own public claims.
There’s a reason it’s hard for outsiders to independently check how often crashes like Mendoza’s happen. A federal Standing General Order requires automakers to report crashes involving Level 2 driver-assist systems, specifically so regulators aren’t relying on manufacturers’ voluntary honesty. But companies are allowed to redact the narrative portion of those reports, and one review of that database found Tesla redacting the narrative in the overwhelming majority of its filings, while GM, Ford, and Toyota rarely do. Combine that with sealed settlements and confidential terms, and the cars generating the most attention-grabbing crashes are also generating the least usable public data about them.
Compare that with how other autonomy companies have handled their own bad days. When one of Waymo’s robotaxis drove into floodwaters, the company pulled a related software update and made the incident public almost immediately. Tesla’s approach to Autopilot’s failures has largely been the opposite: settle, seal, and redact. The regulatory climate isn’t getting friendlier to secrecy, either. States like California are racing to write new rules for autonomous vehicles, including a fight over driverless trucks that has already drawn a legal challenge. The pressure on every automaker to explain, in public, what its automation can and cannot do is only building.
Related Articles
- Court Approves Hertz Deal Widening Rental Options for Drivers Using Hand Controls
- Nissan Called July a “Growth” Story. Its Own Numbers Show China Production Collapsing by Two-Thirds
That gap matters well beyond one courtroom. Insurers price risk using claims history and loss data. Regulators write rules using crash patterns. Attorneys build the next case using discovery from the last one. When the most visible driver-assist system on American roads produces a legal record that’s mostly sealed and a safety record that’s mostly redacted, everyone downstream, including the next owner cross-shopping a used Model S, is working from an incomplete picture of how the system behaves in exactly the scenario that killed Genesis Mendoza: a stopped vehicle, in a lane, in the dark.
Tesla will say a settlement isn’t an admission of anything, and legally, that’s correct. But settle enough fatal cases at the exact moment the same argument is about to reach a jury, and the pattern starts making the argument for you. Companies don’t pay to keep a word out of a courtroom because the word is doing just fine in front of juries. They pay because they already know what a jury would do with it.