@LordCaramac @bwana there's also the issue of how foundation models have been built, even the open-weights ones: the kind of hidden low-pay and psychologically damaging labour for censoring, and how the needed datasets have been collected in unclear or admittedly illegal ways. I would like to have an ethical assessment of professional (and even more, publicly funded) usage of such services and products, even before the legal assessment (see cloning the UX) settles down.