Really?
Perhaps that tweet rather was tempting fate - or perhaps even seen by some as a challenge / something to test?
Though I strongly suspect Musk's "10x" was talking about auto "pilot + driver" ... presumably largely data taken from enthusiasts who will have been paying more attention themselves than the average driver.
My own view is that Tesla drivers using autopilot will likely largely be enthusiasts wanting to tinker and test the technology - like those making youtube videos - so will probably be paying more attention than an average driver, which could skew the statistics somewhat... much of the "10x safer" could be coming from the driver's extra attention while they evaluate their expensive, fancy new toy, not the autopilot itself per se.
I have to admit, watching the Telsa owner's videos on youtube, I really couldn't see Tesla's getting very far without the assistance of a driver. Waymo they certainly ain't!
Telsa's don't seem to be able to select an appropriate speed for the conditions (both speed up and slow down) ... it's funny (if a little concerning) watching the videos and seeing how the drivers frequently 'dial in' a different speed, but don't consider it an intervention.
And other times, the Tesla just keeps waiting and needs the driver to tap the accelerator to tell the car to move away at a junction. And again, the drivers don't see this as being an 'intervention'
So when I read that a Tesla has lost control on a corner with seemingly no-one in the driver seat, I know what my guess would be as to what's gone on.
I note the article I linked above also says...
"But last month, the US National Highway Traffic Safety Administration opened investigations into 27 crashes involving Tesla vehicles."