Cory Doctorow's linkblog · @pluralistic
38964 followers · 35984 posts · Server mamot.fr
Miguel Afonso Caetano · @remixtures
262 followers · 471 posts · Server tldr.nettime.org

: "Many creators have a very different reaction to the emergence of generative ML systems. They worry that their creative work will be exploited by companies building generative ML applications, which in turn will weaken the demand for their work. There are concerns that all the value generated by these tools will go to large tech companies, with none going to the artists and creators whose work is used to train these models. Others are concerned about the commodification of their unique artistic styles[7], or simply want to control how and by whom their work is used. For all of them, the EU approach to TDM /ML opens up an interesting perspective: They can use their ability to opt-out as leverage to set terms and demand compensation.

Given the scale of ML training (which for foundational models is at the scale of the Internet), this lever is unlikely to work very well if used by individual creators on their own. Instead, it seems clear that creators will need to band together to collectively enforce their rights against those who wish to use their works as input for ML training."

openfuture.eu/blog/protecting-

#ai #generativeAI #eu #ml #MachineLearning #copyright #tdm #collectivelicensing

Last updated 2 years ago