TACC is great and definitely a safety feature, though it has recently regressed in that now it thinks the speed limit (on the highway) drops to 25mph almost every time I pass an exit and slams on the brakes, which is unsafe. So let's say that in 2018.50.x TACC is a net win on safety, and in 2019.x so far it seems like a step back.
But Autosteer is a different matter. I think there is one and only one case in which it improves safety, and that is if the driver loses consciousness involuntarily, like a heart attack or something. In this case it has a good chance of bringing the car to a safe stop once it no longer gets steering wheel torque.
In every other case, I believe it reduces safety, even when the driver is alert and paying attention. I use it a lot less now than I used to because I have come to this conclusion. Because it gives people a false sense of security it makes people more likely to quit paying attention, maybe check that text message that just came in, or worse to get in the car and drive when they shouldn't, because they're tired or maybe a bit tipsy. This makes crashes more likely, not less.
Until they are at least L3 (or very very reliable L2), lane keeping systems are unsafe IMO unless they include a very robust driver attention monitoring system. The more mild steering assist and lane departure warning systems are fine.
If nothing else, activating Autosteer and especially NOA seems to make TACC jumpier and more prone to phantom braking, which is why I don't use Autosteer much these days. Tesla, give me rock solid and smooth TACC without phantom braking please before you waste time chasing FSD.
I disagree. I find autosteer useful in stop and go traffic and on long trips. On long trips I do find myself using a mix of full manual control, TACC, and, for a bit of a rest, AP. AP is great when I find myself wandering a bit. This is especially true where lanes have been narrowed to add an additional lane.