These precedents are also creating risk by the way. It hasn't been that long ago since news was posted here of security researchers remotely cutting off car engines or cars repossessing themselves if you don't pay on time. That's the sort of existential risk you're paving the way for by accepting "safety" in exchange for control. It's not really your car anymore, it's the government's car, the insurer's car. What other policies would they like to set, I wonder?
Aren't you notice, my dear safety fan, that the thing you are propagating is effectively adding one more person/actor on the road - a programmer of potentially buggy software, totally not FOSS and without any responsibility for any possible outcomes of possible bugs?
Because nobody has drawn a red line and Ron Jones in 1967 has shown us how vulnerable our freedom is.
We already has a lot of mandatory non-FOSS software in cars which use to phone home which effectively is a streaming (not a video yet, but amount of data is never decreasing). And a lot of so-called safety (surveillance in reality) devices are around us nowadays - phones exposing a lot of uncertain data about us, bank cards which allows banks know where we are right now, etc. If some crackpot politician proves to safety-demanding influencers that video streaming from every moving car helps us to prevent $bad_things_list so that kind of demand is totally possible, especially if using cars for mass killings would become more common than mass shootings.
> I'm just suggesting that regulators should compare accident rates of cars with and without automated breaking.
It's impossible because you have not 100% of the data.
Vehicle regulation should be based on the overall safest option, not 'dignity' or freedom or individualism anything like that.