Interestingly, it's gone the other way in mathematics. It used to be that lots of journals required authors to sign some sort of statement saying that they
wouldn't post PDFs on their web pages. (I remember one professor who had a page essentially saying "Here's the PDFs. Sue me, Elsevier!" Unfortunately, I don't remember who it was) Eventually, I think in response to the massive, overwhelming popularity of the arXiv, which I think most mathematicians would choose over journals if forced to choose, the big publishers decided, slowly, piecemeal, and behind the times as always [0], to go with the times and stop fighting something that was going to happen anyway.
[0] With an exception for the AMS, which has always (as long as I have cared, and checked, anyway) had publication policies that are as author- and reader-friendly as one can probably expect in the real world.