After reading about this for a while, just now, I'm much less comfortable with the idea of trusting an automotive autopilot system. Just imagine if your taxi driver told you "I will drive perfectly safely, except if there is a truck parked in the road. Then I will run into it at speed." It seems like such a glaring omission. If I were an engineer I don't think I would be comfortable releasing an autopilot system with this kind of safety issue.

Which is why most companies aren't willing to call these things "autopilots" - Mobileye, for example, pitches its systems as assistants which will catch some of the driver's mistakes, similar to the "collision avoidance" language. This also comes with the corollary that they often aren't willing to give these systems as much control of the car as Tesla does.