Here is the contradiction most people miss: the wrong-way robotaxi in Tampa is the safest thing Tesla did that month.
Not because an empty Model Y reportedly stopped dead facing the wrong direction on a one-way street is fine — it isn't. Drivers reportedly swerved around it on North Armenia Avenue at West Cypress, no one behind the wheel, a single day after the Tampa launch. The dashcam clip went everywhere, and it deserved to.
But that footage is a driving error you can see. It's discrete, local, and recoverable — reportedly a bug in a small service area covering parts of West Tampa and Tampa Heights.
The wrong-way driving that actually threatens Tesla isn't happening on a Tampa street. It's happening on paper, in letters to European regulators, where nobody is filming.
The wrong-way driving that actually threatens Tesla isn't happening on a Tampa street. It's happening on paper, in letters to European regulators, where nobody is filming.
Here is what Tesla wrote down when it thought only regulators were reading.
In a slide presentation sent to Swedish authorities, Tesla policy manager Ivan Komusanac claimed Full Self-Driving could have "saved 32,000 lives" and prevented 1.9 million injuries, per the Reuters investigation reported by Electrek. That's the headline number. It's also, on inspection, a specific engineered fiction.
The 32,000 figure assumes every vehicle in the United States — every freight truck, every motorcycle, every 20-year-old pickup — is replaced by an FSD-equipped Tesla. It's not a projection. It's a fantasy dressed as arithmetic.
And it sits on top of a deeper manipulation, one that's easy to state and hard to unsee once you have.
Tesla's core "10 times safer than human drivers" claim rests on comparing two things that should never be compared. It reportedly measures crashes in FSD-equipped Teslas severe enough to trigger an airbag deployment. Then it sets that against the crash rate for all U.S. vehicles — including every minor fender-bender, every parking-lot scrape, every crash too small to pop an airbag.
Airbag crashes are, per the Reuters investigation reported by Electrek, a small, severe subset of all crashes. Comparing that subset to the total, per the Reuters investigation reported by Electrek, inflates the safety margin by a factor of three. You are counting only Tesla's worst crashes and pretending they represent all of them, then racing that against everyone else's every crash.
That's not a rounding error. That's a designed comparison, chosen because it produces the number Tesla wants.
There's a second thumb on the scale. Tesla benchmarks against the average U.S. vehicle, which is about 12 years old.
Newer cars from every manufacturer crash less, because they have better structural design, better braking, better sensors. A new Toyota Camry also beats the average U.S. vehicle on this metric — and nobody thinks Toyota has solved autonomy. Tesla is partly measuring the safety of being a new car, and calling it the safety of FSD.
Stack these together: the airbag-versus-all-crashes trick (roughly 3x), the geriatric-fleet baseline, and then the everyone-drives-a-Tesla extrapolation. Each layer multiplies the last. By the time you reach "32,000 lives," you're several fabrications deep into a number presented as science.
This isn't a fringe reading. Ten of the eleven independent traffic-safety researchers who reviewed Tesla's methodology called it misleading marketing, not a serious safety study, per the Reuters investigation reported by Electrek.
Ten of eleven. That's not a debate. That's a consensus with one holdout.
Now here's why the paper matters more than the pavement.
For years, Musk has posted inflated safety stats on social media. That's background noise at this point — a CEO puffing his product. Annoying, discountable, legally cheap.
What changed is the venue. Tesla took that same self-published math and submitted it directly to government regulators as part of an official approval process. The marketing became a filing.
In a November 2024 letter to RDW, the Dutch road authority, Tesla linked its safety report and asserted that "increased usage" of FSD "leads to safer roads." After more than a year of testing, RDW approved FSD in the Netherlands in April — and is now carrying Tesla's bid for EU-wide approval.
RDW says it "does not rely on marketing claims or external statistics" and runs its own tests. Maybe so. But when Reuters asked whether RDW had actually assessed the validity of Tesla's U.S. safety figures, the agency didn't say.
That silence is the whole game. This publication has a standing question: do regulators get independent access to operating logs, or do they trust automaker-supplied data? Here the answer arrived in the ugliest possible form — Tesla handed over self-produced math, and at least one approving regulator won't say whether it checked it.
Contrast that with Norway, which refused to be polite about it. When Tesla enthusiasts emailed the Norwegian Public Roads Administration urging approval, Stein-Helge Mundal responded that Tesla's figures "are self-produced," making it "difficult to find correlation with the authorities' accident statistics."
That's a regulator naming the exact defect: the numbers come from the company that benefits from them, and they don't line up with independent crash data. Norway saw a marketing document and called it one.
The European Transport Safety Council put the fix plainly. Its spokesperson Dudley Curtis said that if Tesla wants to make safety claims, it should "give the data to a university, have it independently verified by a qualified researcher, and then let's talk."
That is the entire standard, stated in one sentence. Tesla has not met it. Instead it routed unverified numbers through the country most willing to move fast.
And this is the closed loop. A discredited figure — flagged by ten of eleven researchers, flatly called self-produced by Norway — is being laundered into law through the Netherlands, which now speaks for Tesla at the EU level. The Dutch approval doesn't refute the criticism. It bypasses it.
With the Netherlands, Lithuania, Belgium, and Denmark having already approved nationally, and once 55% of member states representing 65% of the population vote yes, the fantasy math stops being a talking point. It becomes the regulatory basis for putting the system on public roads across a continent.
There's a reason Tesla is pushing this hard right now. FSD approval in Europe is central to regaining market share in a region where sales cratered and where BYD has now outsold Tesla for multiple consecutive months. The company needs the win. That's the pressure behind the paperwork.
Which brings us back to the ground, where these numbers cash out.
In Katy, Texas, a Tesla Model 3 reportedly running FSD veered out of its lane and crashed into a home. It killed 76-year-old Martha Avila.
Musk's response was that the crash "makes no sense" because FSD drives slowly in residential areas. Tesla's autopilot VP Ashok Elluswamy said the driver overrode the system by flooring the accelerator to 73 mph. Those claims may be true, partly true, or convenient — and the point is that we largely have to take Tesla's word, because the company has a documented record of making crash data hard for authorities and attorneys to obtain.
That's the same problem as the European filings, wearing different clothes. In both cases the safety story is authored by the party with the most to lose from an honest one.
The wrong-way robotaxi will be patched. Somebody will push an update, the North Armenia Avenue clip will fade, and Tampa will move on.
The 32,000-lives number won't be patched. It's already inside the approval process, doing exactly what it was built to do.
Watch the paper, not the pavement. The dangerous wrong-way driving is the kind that ends up in a regulator's inbox, stamped, and voted into law — with a dead 76-year-old in Katy as the reminder of what's downstream.