You should be safe and free to brake without deep introspective thought, without fear of causing an accident. The real world is not that simple but that's how the law is generally setup.
And as a further point they'd probably say that if that person following had such a system, there's also less chance they'd rear-end you in turn. I was wary of the systems in my own car, but after 6 months of driving they are far more reliable than I'd feared and I'm quite impressed.
The kinds of mistakes described elsewhere in this thread have not been of the form "system mistakenly brakes my car and someone behind me rear-ends me because they were following too close". So your comments, while valid for that particular scenario, do not appear to be relevant to what other posters are concerned about.
Car reviews are largely useless fluff - a real analysis of the differences between brands for this kind of thing could actually be useful.
That's because those other comments all assume that "automatic emergency braking" will only do exactly what it's supposed to do and will never make a mistake. But any sensible person knows that's not going to happen. These systems will make mistakes, and some of those mistakes will cause harm. So you can't just make a blanket claim that such systems are good. You have to actually look at the data and balance the harm they avoid by doing what they're supposed to do when they're supposed to do it, against the harm they cause when they make mistakes.
The obvious next question is, does the proposed rulemaking here consider the consequences of the system making mistakes? As far as I can tell, the answer is no, with just one qualifier. The claims about crashes prevented and lives saved are not net savings, accounting for harm done by the system making mistakes; those claimed numbers assume the systems will always perform perfectly and will never cause any harm, only prevent it. The one qualifier is that the proposal does include two "false positive" tests to try to spot systems that brake when they're not supposed to. But there is no analysis given to show why those particular false positive tests were picked or why they would justify the extremely high level of confidence in the systems not making errors that the rest of the proposal shows.