Tesla Swore the Fatal Autopilot Data Was Gone. A $15K Bug-Bounty Hacker Found It in Minutes — and It Cost Them $243 Million

Tesla Insisted the Fatal Crash Data Was Gone. A $15K Bug-Bounty Hacker Found It in Minutes—and It Cost the Company $243 Million

Picture this: a dark Florida night in 2019. A Tesla Model S glides along a Key Largo road with Autopilot engaged. The driver, George McGee, drops his phone, glances down for a second—and the car keeps going. It blows through a stop sign and a flashing red light at more than 60 mph, slamming into a parked Chevy Tahoe.

Standing beside that Tahoe are 22-year-old Naibel Benavides Leon and her boyfriend, Dillon Angulo. She dies. He is critically injured.

For years afterward, the grieving family and Angulo fought for answers. The most important piece of evidence—the “collision snapshot” that Tesla vehicles automatically capture and send to company servers in the final seconds before a crash—seemed to have vanished. Tesla said it didn’t have the full file. A service technician called the data on the car’s Autopilot computer corrupted. Case closed, right? Wrong.

The Man Who Refused to Take “Missing” for an Answer

Enter a quiet, ponytailed security researcher known online as @greentheonly—Green, for short. He’d already worked with Tesla through its bug-bounty program, earning $15,000 for each software vulnerability he reported. He knew the company’s systems inside and out. When the plaintiffs finally recovered the original Autopilot hardware from the Florida Highway Patrol in 2024, they called him.

They met at a Starbucks near Miami International Airport. Green ordered a Venti hot chocolate, opened his ThinkPad, and plugged in a forensic copy of the crashed car’s computer. Within minutes—some accounts say 15—he found exactly what Tesla had insisted was gone: the complete collision snapshot, complete with video of the final moments and a transmission ID proving the data had been successfully uploaded to Tesla’s servers right after the impact.

The local copy on the vehicle had been marked for deletion, standard procedure. But the information was still there. And it told a damning story: the system had detected the end of the road, the parked Tahoe, the pedestrians, the stop sign, and the flashing light… and continued straight toward them.

“For any reasonable person, it was obvious the data was there,” Green later said.

Suddenly Tesla “found” its own copy on its servers. A company lawyer called the earlier handling “clumsy” and admitted, “We didn’t think we had it, and we found out we did.” The judge saw no deliberate cover-up, but the damage was done. The data that was never supposed to exist became the centerpiece of the trial.

The Jury Speaks

In August 2025, a Miami federal jury assigned Tesla 33 percent of the blame and the driver the rest. They awarded roughly $129 million in compensatory damages (Tesla’s share about $43 million) plus a staggering $200 million in punitive damages. Total bill for Tesla: approximately $243 million.

It was the largest verdict ever handed down against the company and the first time a jury held Tesla liable in a wrongful-death case tied directly to Autopilot. Plaintiffs argued the system was defective, driver monitoring was inadequate, and marketing had oversold what the technology could actually do. Tesla insisted the crash was entirely the driver’s fault and vowed to appeal. A federal judge upheld the verdict in early 2026; the fight continues in higher courts.

Why This Story Matters

This wasn’t just about one crash or one missing file. It was about who controls the digital truth when a connected car is involved in a tragedy—and what happens when that control is challenged by someone who knows the systems better than the company hoped.

A $15,000-a-bug white-hat researcher sipping hot chocolate in a Starbucks rewrote the narrative. The data Tesla said didn’t exist not only existed—it helped deliver a historic $243 million judgment.In the age of Autopilot, Full Self-Driving, and the coming robotaxi era, that is a warning no automaker can afford to ignore.