James Grimmelmann (University of Maryland Francis King Carey School of Law) has posted There's No Such Thing as a Computer-Authored Work – And It's a Good Thing, too (Columbia Journal of Law & the Arts, Forthcoming) on SSRN. Here is the abstract:
Treating computers as authors for copyright purposes is a non-solution to a non-problem. It is a non-solution because unless and until computer programs can qualify as persons in life and law, it does no practical good to call them "authors" when someone else will end up owning the copyright anyway. And it responds to a non-problem because there is nothing actually distinctive about computer-generated works.
There are five plausible ways in which computer-generated works might be considered meaningfully different from human-generated works: (1) they are embedded in digital copies, (2) people create them using computers rather than by hand, (3) programs can generate them algorithmically, (4) programmers as well as users contribute to them, and (5) programs can generate them randomly. But all of these distinctions are spurious. Old-fashioned pen-and-paper works raise all of the same issues. A close look at the creative process shows how little really changes when authors use digital tools. The problems posed for copyright by computer-generated works are not easy, but they are also not new.