I've people on previous teams that thought this, and their comments were consistently the ones that were the most out-of-date because they never valued fixing them. If you value comment accuracy during PR reviews then you're less likely to end up in that situation in the first place.

This. I totally agree.

One fallacy I've seen is a reviewer only looking at the diff on Github, where a minor change in the middle of a function is made, and they don't expand the diff to notice that the new change completely invalidates a comment, params, call signature, return value..etc.