A lawsuit filed by several authors against Meta centers on Meta’s alleged use of pirated books for AI training data and the technical details of BitTorrent which was used to obtain them. Yesterday, Meta filed a motion for summary judgment, while countering the authors’ request to resolve the copyright claims in their favor. Meta’s request includes new information, including the revelation that its uploads of ‘pirate’ library data were roughly 30% of the data it downloaded.
Using upload stats from AWS is ridiculous. Bittorrent is heavy on the upload and requires more than 10% of your download bandwidth in uploads just for protocol overhead on a small number of peers without even considering seeding
More than 10% but less than 30%? They initially said that they only downloaded, but now their argument seems to be they uploaded little enough that no-one got a complete copy of anything from them.
Likely, wonder if they uploaded much of the data from Amazon to themselves?