Those claiming AI training on copyrighted works is “theft” misunderstand key aspects of copyright law and AI technology. Copyright protects specific expressions of ideas, not the ideas themselves. When AI systems ingest copyrighted works, they’re extracting general patterns and concepts - the “Bob Dylan-ness” or “Hemingway-ness” - not copying specific text or images.
This process is akin to how humans learn by reading widely and absorbing styles and techniques, rather than memorizing and reproducing exact passages. The AI discards the original text, keeping only abstract representations in “vector space”. When generating new content, the AI isn’t recreating copyrighted works, but producing new expressions inspired by the concepts it’s learned.
This is fundamentally different from copying a book or song. It’s more like the long-standing artistic tradition of being influenced by others’ work. The law has always recognized that ideas themselves can’t be owned - only particular expressions of them.
Moreover, there’s precedent for this kind of use being considered “transformative” and thus fair use. The Google Books project, which scanned millions of books to create a searchable index, was ruled legal despite protests from authors and publishers. AI training is arguably even more transformative.
While it’s understandable that creators feel uneasy about this new technology, labeling it “theft” is both legally and technically inaccurate. We may need new ways to support and compensate creators in the AI age, but that doesn’t make the current use of copyrighted works for AI training illegal or unethical.
For those interested, this argument is nicely laid out by Damien Riehl in FLOSS Weekly episode 744. https://twit.tv/shows/floss-weekly/episodes/744
Not to fully argue against your point, but I do want to push back on the citations bit. Given the way an LLM is trained, it’s not really close to equivalent to me citing papers researched for a paper. That would be more akin to asking me to cite every piece of written or verbal media I’ve ever encountered as they all contributed in some small way to way that the words were formulated here.
Now, if specific data were injected into the prompt, or maybe if it was fine-tuned on a small subset of highly specific data, I would agree those should be cited as they are being accessed more verbatim. The whole “magic” of LLMs was that it needed to cross a threshold of data, combined with the attentional mechanism, and then the network was pretty suddenly able to maintain coherent sentences structure. It was only with loads of varied data from many different sources that this really emerged.