There are approximately 250 million cars on the road and they only used data from 8 million? That's 3% of cars on the road to extrapolate into all the cars on the road. Seems like a huge flaw especially since we didn't know how they got that subset. All seems like click bait as most articles related to Tesla are...
Another good way / better way to see what cars are dangerous are insurance rates. Since insurance companies take in way more data than 8M cars when determining rates.