[Publishers claim sites like Sci-Hub] “have no incentive to ensure the accuracy of scientific articles, no incentive to ensure published papers meet ethical standards, and no incentive to retract or correct articles if issues arise.”
The correct response to this should be to always note in citations when a paper was accessed through sci-hub. Then the reader is warned, and can compare against the canonical version if they are worried.
Sci-Hub is a library, a repository, not a publisher. The original article publisher has editorial responsibility for article selection, not Sci-Hub, just as with any repository.
The accuracy of articles themselves and the integrity of them is trivial to establish as SciHub, LibGen, and ZLibrary all index principally on checksums of files. If there's a modification, the checksums will reveal which items are affected.
Note the MD5 hash provided for Stephen Breyer's seminal article "The Uneasy Case for Copyright" (1970): 8F8B57972236B04621795C49A15E8C2E
[Publishers claim sites like Sci-Hub] “have no incentive to ensure the accuracy of scientific articles, no incentive to ensure published papers meet ethical standards, and no incentive to retract or correct articles if issues arise.”
The correct response to this should be to always note in citations when a paper was accessed through sci-hub. Then the reader is warned, and can compare against the canonical version if they are worried.