Is it just me, or is the training behavior of all of the big AI companies just another attempt to put knowledge behind a paywall in much the same way as Aaron Swartz described in the Guerrilla Open Access Manifesto?