I was watching a report about the Tesla murder of a woman in Texas, and this chart popped up.

This is Tesla’s Vehicle Safety Report rebroadcast without a single control applied. CBS intentionally, openly, runs a fraudulent “safety” graphic claiming roughly 8x safer (5.5M ÷ 660K = 8.3, 1.6M ÷ 222K = 7.2) in a story about Tesla killing a woman, directly above a chyron saying as much.

The graphic asserts the exact inverse of the news it runs with, a perfect illustration of targeted disinformation. The Tesla numbers are inflated at both ends.

Numerator suppressed. Tesla counts a crash only inside roughly five seconds of disengagement where NHTSA’s reporting order specifies thirty, and counts mainly events at the airbag and restraint threshold. By the agency’s own finding Tesla captures data on around 18 percent of police-reported crashes. Fewer crashes counted means more miles per crash as an intentionally artificial construction.

The Tesla death headline is a cooked definition, not a measurement. It’s Enron, it’s WorldCom, it’s Bernie Madoff.

Denominator gamed. The 5.5M figure is supervised, highway-weighted miles in good conditions. The “US average” is every road, every condition, every vehicle age, including cars built before electronic stability control. Another artificial construction to lie about safety. New beats old carries no information about the system.

And their “active supervision” label is propaganda that concedes the rest: a human monitor was preventing crashes, so the number measures human plus machine, then it credits the unsafe machine instead of the actual safety from a human intervention.

Closed and unsafe. Singer testified there is no math and no science behind the Vehicle Safety Report. CBS ran the lie.

Waymo adjusts for road and neighborhood type, compares against human drivers in the same markets, and publishes through outside review; Tesla keeps the data secret and seeks none.

A self-attesting number, a lie, against an externally validated one. Run the apples-to-apples correction and the advantage collapses. Marco Benedetti matched airbag to airbag and got about three times, calling even that generous because Tesla measures a Tesla driver against the average driver and hides the rest behind fleet age. Three times worse, generously. The Tesla chart claims eight times better.

Here is the cleanest way to state the fraud. The latest 8x worse data from Tesla robotaxis is the same category of driving the CBS chart is bragging about: supervised autonomy with a monitor in the seat. Against NHTSA’s police-reported baseline of roughly one crash per 500,000 miles, the supervised fleet runs about eight times the human rate. On the tighter baseline the arithmetic is 7 crashes in roughly 300,000 miles against one per 700,000, which is 16.3x. Same multiplier, exact opposite result.

CBS broadcasts the fraudulent 8x safer slide for the exact driving mode that measures 8x more dangerous, once a real baseline is used. The two numbers describe the same thing and differ by a factor near sixty.

Another external check also proves the lie. LendingTree’s analysis of 30 brands put Tesla drivers first in accidents at 23.54 per 1,000. Fatal rate runs 5.6 deaths per billion miles against 2.8 for all brands. The marketing chart is a bald faced lie, which begs why a television segment ran it unedited instead of asking me. Someday, maybe.

Imagine police chasing a robot and yelling STOP at it, while it tries to flee with humans held hostage. Why doesn’t the robot stop when ordered? Or more to the point why isn’t there a stop button that police remotely activate?

Waymo keeps turning people into hostages, scaring customers so badly they never want to get in one again.

Records show there were seven cases on Bay Area roads just in one day last month.

In a statement sent to San Francisco ABC affiliate KGO, Waymo says they’ve identified improvement areas for their cars in construction zones.

They’ve also voluntarily restricted freeway access while they implement those improvements.

It’s a move supported by Slade, who says he and his fiancé even feared for their lives.

“In that moment it’s like, ‘oh this technology is not ready. This is 100% not ready.’ If something else had gone wrong, someone in that road might have got hit. We might have crashed in the car,” Slade said.

He’s not wrong.

“There were signs. There were lights. There were cones. And it went through the cones and then sped up straight away,” Slade said.

To make matters even worse, Slade said a nearby highway patrol car began chasing them after seeing what was going on.

“Shouting, ‘Stop Waymo. Stop Waymo. Stop Waymo.’ We’re like- what the heck is going on?” Slade said.

Shouting stop should work. In fact, anyone on the street should be able to shout stop at a Waymo anytime, just like a human driver.

Seven failures a day isn’t sustainable. The robots under the banner of AI are ignoring all their warnings, ignoring human commands, and rushing towards increasing public danger. It literally bypasses guardrails.

There’s a long history of Tesla drivers saying they didn’t touch the accelerator, and Tesla saying that it was constant, pegged at 100%. The pattern is back, with a string of five crashes just this month.

Ashok Elluswamy, Vice President of Auto-Pilot for Tesla, commented on the X social media platform about the crash. In a reply complaining… “…accelerator all the way to 100% of the accelerator pedal in this residential area. They reached a speed of 73 mph during the crash, and had the accelerator pressed even after the crash.”

That statement itself indicates Tesla failed at multiple levels, and carries direct liability for the death in Texas.

Notably, humans have to work very hard to peg an accelerator at 100% through a building, especially after the crash. They pump, they panic, they alter, they move. A crash force would make it almost impossible to maintain pedal pressure. To be stuck at a constant indicates the defect is the software getting bad signal, which also should have an integrity check and override.

Detecting 100% pegged acceleration in a residential area without room for that mode is a trivial failsafe to design. The car should automatically refuse.

Five of these crashes into buildings in about the last 30 days, by AI supposedly able to “see” things to avoid, already begs the question why everything “X” of Elon Musk is so obviously defective. Integrity breaches, especially fatal ones, should have accountability that goes straight to him.

Tesla fraudulently sold these cars for a decade as perceiving the world with cameras and neural nets. If the perception stack was running, the car had every input needed to know it was on a residential street closing too fast on a fixed structure at short range. Automatic emergency braking is supposed to arbitrate exactly that case, and Tesla has long claimed its AEB functions even under driver throttle.

Elluswamy’s account forks into two branches, both of which confess Tesla’s deadly failures.

Either sustained full pedal was allowed to suppress AEB by design, meaning the override was deliberately disabled, or perception failed to register a brick house dead ahead, meaning the system failed at the one task it is marketed to perform. Invoking the pedal concedes that nothing in the car arbitrated a full-throttle command against its own perception.

The Tesla defense is self-incriminating.

Drive-by-wire pedals carry dual redundant position sensors with different output slopes, and they do cross-check: do the two sensors agree? Two channels both reading 100% pass, while neither answers the second layer, command plausibility. Should a car apply and hold maximum torque while its own cameras see a structure fifteen meters away in a 25 mph zone, while its positioning system contradicts the path? The sensor integrity check validates the hardware. Nothing validates the command against context. That is the failsafe of perception and powertrain never cross-checked, let alone many other options.

The Tesla executive read out his opaque proprietary telemetry on social media as if he’s God himself delivering judgment, before the investigation, with no third-party verification, from a bully pulpit the Washington Post has documented as repeatedly “losing” and withholding EDR data from litigants.