P. K. G. Williams’s preprint treats scientific literature and open-source software as collectively accumulated information resources, comparing AI commercialization and low-quality generated submissions with historical enclosure. The argument is that review burdens and diminished motivation could weaken production communities even if the underlying materials remain accessible. arXiv has in fact imposed limits of two submissions per submitter per month and three active submissions from October. That is evidence of operational pressure, not causal proof that AI must destroy science; the paper also acknowledges the possibility of useful AI contributions.