Tesla allegedly hid fatal accidents to keep autonomous tests on the road

April 20, 2026
Explore the elegant interior of a Tesla Model X captured in a Sydney park setting.
Photo by Raimundo Campbell on Pexels

A leaked trove lifts the veil

It has been reported that internal Tesla documents reveal the company concealed thousands of incidents tied to its Autopilot and other driver-assist systems. The leak, published by Swiss broadcaster RTS and other outlets, allegedly shows more than 2,400 complaints of unintended acceleration and upwards of 1,000 crashes — many marked as "unresolved." Shocking? Yes. Surprising? Not entirely; the auto industry has been racing to ship advanced driver assistance while regulators try to catch up.

Hallucinations on the highway

Engineers call them "hallucinations" — the AI sees things that aren't there, or misses things that are. On the road, those mistakes can be lethal. It has been reported that some Teslas brake or accelerate suddenly with no clear trigger, and in multiple cases the system detected obstacles without taking effective evasive action. Think of it like a misfiring autopilot that reads the world wrong — a digital blind spot that can kill.

Victims, buried data and a jury’s answer

Families say the human cost has been enormous. Dillon Angulo, who survived a crash that killed his partner Naibel Benavides, told investigators he felt like a "guinea pig" when he learned the car had been in Autopilot. It has been reported that data from the vehicle’s "black box" was initially described by Tesla as corrupted; experts later recovered files that allegedly showed the company knew about failures very shortly after some crashes. A jury has since ordered Tesla to pay more than $243 million in damages — a landmark verdict in Autopilot litigation.

Bigger questions remain

So what now? Regulators, courts and consumers are asking a blunt question: should commercial roads be used as rolling testbeds for emerging AI? The case is both a legal turning point and a cultural one — a cautionary tale for any company that treats machine learning like product polish rather than a safety-critical system. Expect more scrutiny, more lawsuits, and a heated debate about how fast industry should move when lives are on the line.

Sources: rts.ch, Hacker News