No, that was never the case. Anthropic got told in court that using copyrighted works that they had legally procured for model training was fair use, but since they had used a pirated data set, they got ruled against - there’s no fair use protection for things you haven’t actually bought. Subsequently, they settled that lawsuit for $1.5B
No, that was never the case. Anthropic got told in court that using copyrighted works that they had legally procured for model training was fair use, but since they had used a pirated data set, they got ruled against - there’s no fair use protection for things you haven’t actually bought. Subsequently, they settled that lawsuit for $1.5B
Yeah, thanks for clarifying. I was more saying it with tongue in cheek as a jab at meta’s bittorrenting books, which even itself isn’t settled.