Thanos Apollo <[email protected]> writes: > "pelzflorian (Florian Pelz)" <[email protected]> writes: > >> I doubt either case is how the LLMs’ effect on copyright will be. >> Traditional copyright will likely be affirmed and extended. Especially >> since fair use is no concept in European copyright, where exceptions for >> non-commercial use are the alternative. > > This misreads EU copyright. There is no "fair use," true; but Article 4 > of the CDSM Directive is an explicitly commercial TDM exception, open to > for-profit AI training with rightsholder opt-out as the only check. On > training, the EU is structurally more permissive than the US, which has > no statutory TDM exception at all.
Yes there are more exceptions unrelated to non-commercial like pastiche art. Yes, that commercial GPT training is lots of correlations under the (very pertinent) TDM exception is alluded to in law. But the EU analyses Ludo shared in the GCD do acknowledge authors and current WTO rules rank higher than the very saddening government recommendations on schools that I did not know. Thanks for sharing. > [3] > https://eurydice.eacea.ec.europa.eu/news/germany-recommendation-action-use-artificial-intelligence-school-education-processes > [4] > https://kmk.org/fileadmin/Dateien/veroeffentlichungen_beschluesse/2024/2024_10_10-Handlungsempfehlung-KI.pdf How sad. It goes against academics’ well-founded reasoning (for example, Monika Oertner’s) for a ban on LLMs in educating university students when not specifically educating on how to LLMs. Regards, Florian
