On 4/30/25 07:49, Steffen Wolfrum wrote:
Hi,
we just saw the Accessibility Directive, now there is another demand: the TDM protocol (concerning the reuse of protected works for AI training purposes). Hi Steffen,
I have no doubt your suggestion or request is well intended. But do you really think this may be worth the hassle? I mean, from Llama is suspected that it has been trained with more than 80TB of unauthorised copies of books (from shadow libraries https://en.wikipedia.org/wiki/Anna%27s_Archive#Meta_lawsuit, https://en.wikipedia.org/wiki/Library_Genesis#Alleged_Meta_AI_training). Do you really think that reservations will be honoured? Even when they honour reservations, research purposes seem to be granted without possible reservations (probably also to commercial entities). As for the record, Google was taken to court in Spain by an individual because of not complying with their robots.txt (not allowing Google to crawl their content). This went to the Spanish Supreme Court, who ruled that this was a case of «ius usus innocui» (right to harmless use; not exactly fair right). This was a commercial entity (well, an emporium), making a full copy of a copyrighted work and the author has (or had before this ruling) the right to control distribution of their work. Sorry for being straight, but this is similar to data protection. As long as we give our data (just by using your phones or using internet), our data are collected, analyzed and traded in order to influence our behaviour. I’m afraid that this is part of sharing documents on the internet. Just my two cents, Pablo