This is fear mongering disguised as concern. Tesla publishes data about incident rates for drivers both on and off autopilot and they are markedly lower than other drivers.
What if you control for demographic factors? I'd expect that new drivers and the elderly have accidents at a much higher rate than the average driver and also have low Tesla ownership rates.
Edit: I see they do have base rate w/o autopilot listed and it does significantly differ from national average.
Does it distinguish between the types of road driven on? It's a very important distinction because where people sensible people choose to activate the system is probably on long journeys where crashes are already rare.
https://www.tesla.com/VehicleSafetyReport