On June 22, 2026, a Tesla crashed into a residential swimming pool and home. The details of that particular crash—whether the car's self-driving system was on, where it happened, the driver's condition—haven't been publicly confirmed yet. But the incident highlights something bigger: what does Tesla actually know about its self-driving cars, and how much does it tell the public?

Two Different Records

Tesla keeps two separate sets of information about how its cars drive. The first is a public safety report that Tesla publishes every three months. It contains big-picture numbers: how many miles Tesla's self-driving system drove that quarter, what kinds of roads (highways, city streets, residential areas), and whether the system or the driver was in control. It's like a report card for the whole fleet.

The second is a black box called an Event Data Recorder (EDR). It sits in every Tesla and captures detailed information before, during, and after a crash: did the airbags deploy, did the car detect obstacles, was the self-driving system on. Think of it as a security camera for the car. Only investigators, police, the car owner, and courts can access this data—not the general public.

The key difference: the public report tells you if Tesla's system is generally safe. The black box tells you what happened in a specific crash.

Who Gets to See What

When there's an accident, investigators and regulators like the National Highway Traffic Safety Administration (NHTSA) can request the detailed black box data. Federal law allows them to do that. If Tesla's self-driving system was turned on at any point during the crash sequence, Tesla is required by law to report it to NHTSA.

The problem is timing. The detailed information lives inside Tesla's computers and only comes out through legal requests or if the car owner asks for it themselves. That can slow down official investigations.

Why the Public Report Isn't Enough

Tesla's published safety numbers compare its self-driving cars to national driving statistics. But the comparison has a built-in weakness: the miles driven by Tesla's self-driving system come from experienced users, mostly on highways with clear markings and good weather. The national average includes all drivers in all conditions—bad weather, heavy traffic, inexperienced drivers. They're not comparing the same thing.

Also, a fleet-wide safety report can't explain what went wrong in one specific crash. Even if the numbers look good overall, something else could have caused this accident—the driver wasn't paying attention, the system disengaged just before impact, or something unexpected happened on that particular street.

For people investigating crashes and deciding who's responsible, the detailed black box data is what matters. For the public trying to understand whether self-driving cars are safe enough to drive on residential streets, all we really get is the aggregate numbers Tesla chooses to publish.

Right now, that's a gap nobody has fully solved. Regulators have the tools to investigate individual crashes, but the public has only the overall statistics. Each time something like this happens, it brings the question back into focus—what should we know, and how much of this data should be transparent?