I think one of the replies is also really informative:
To abuse this property you need to get the state of the hash to match a state you get when running the decryption of the blockcipher underlying the compression function. Finding such a match requires a meet-in-the-middle attack with cost $2^{n/2}$ and thus isn’t cheaper than finding a collision.
[Comment removed by author]
I think one of the replies is also really informative:
I feel like an “inaccurate” tag would be in order.
It feels slightly odd that there’s an “incorrect” downvote for posts, but no “incorrect” flag for articles.
I’d add it if it existed :D