To help train AI models, Meta and other tech companies have downloaded and shared pirated books via BitTorrent from Anna’s Archive and other shadow libraries. In an ongoing lawsuit, Meta now argues that uploading pirated books to strangers via BitTorrent qualifies as fair use. The company also stresses that the data helped establish U.S. global leadership in AI.
I hear you, but hear me out… They’re creating products from the consumed torrents, which absolutely contained copyrighted materials. I’m not trying to capitalize my torrents. Although, I did use cracked photoshop back in high school for a $200 job.
And to be completely honest with you, I don’t really care about copyright infringement so much, after it’s become a tool for organizations like Disney or whoever to abuse as they please. But the main body of work torrented here would be corpus’ of text, music, … a lot of stuff that independent producers created and rely on for income.
I found this particular video quite insightful on the impact within the music industry: https://youtu.be/QVXfcIb3OKo
To be fair to Meta, I’d have to say that I don’t really know what models they’re training via that data and how they’re using the resulting products. This is Meta, though, a pioneer and industry leader in the process of surveillance capitalism. I don’t particularly have high expectations for them.