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
Operating system have been used to commit copyright infringement much more effectively and massively by copying copyrighted material verbatim.
OS vendors are not liable, the people who make and distribute the copies are. The same applies for Word processors, image editors etc.
You are for a massive expansion on the scope of copyright limiting the freedoms of the general public not just AI corps or tech corps.
Using your logic, the one making the copy in a word processor is the person typing, and the one making the copy in this LLM is OpenAI
Nope. The output is based on the users input in both cases.
No, the output in a word processor is explicitly created by the user, whereas the output created by a LLM is based on the training data OpenSI scraped and influenced by a user prompt
You need a very specific prompt to make a copy. Even to just be similar enough you have to put the proper input and try a lot of repetitions.
That’s why the right holders are going after the training which included copying by the AI corpos.
In your dream land right holders could just prompt the AI till it spit something close to their work and sue the AI corp for that. Repeat as needed ; infinite money glitch.
Obviously it doesn’t work that way.
OS vendors aren’t selling¹ what users copy into the clipboard.
¹ Well, Microsoft probably is, especially with that recall bullshit, and I don’t trust Google and Apple not to do it either… but if any of them is doing it they should get fined into bankruptcy.
Neither are AI vendors. We have locally hosted AI models and they don’t contain what they output. You can tell by the actual size.
Those analogies don’t make any sense.
Anyway, as a publisher, if I cannot get OpenAI/ChatGPT to sign an indemnity agreement where they are at fault for plagiarism then their tool is effectively useless because it is really hard to determine something in not plagiarism. That makes ChatGPT pretty sus to use for creatives. So who is going to pay for it?
Yes they do.
Which is why you want an agreement to make them liable for copyright infringement (plagiarism is not a crime itself).
You would have to pay for distributing copyright infringing material whether created by AI or humans or just straight up copied.
I don’t care if AI will be used,commercially or otherwise.
I am worried about further limitations being placed upon the general public (not “creatives”/publishers/AI corps) either by reinterpretation of existing laws, amendment of existing laws or legislation of brand new rights (for copyright holders/creators, not the general public).
I don’t even care who wins, the “creatives” or tech/AI, just that we don’t get further shafted.
Something like Microsoft Word or Paint is not generative.
It is standard for publishers to make indemnity agreements with creatives who produce for them, because like I said, it’s kinda difficult to prove plagiarism in the negative so a publisher doesn’t want to take the risk of distributing works where originality cannot be verified.
I’m not arguing that we should change any laws, just that people should not use these tools for commercial purposes if the producers of these tools will not take liability, because if they refuse to do so their tools are very risky to use.
I don’t see how my position affects the general public not using these tools, it’s purely about the relationship between creatives and publishers using AI tools and what they should expect and demand.
“Generative” is not a thing in copyright law.
You regard them as different to tools like Word. That does not exist in the law.
When you originally posted that they OpenAI should be on the hook I thought you meant they were the ones commiting copyright infringement. Not that they would violate private contracts with their customers.
Private agreements is not my business.
There is however a push by both sides to settle this in law. Whatever happens will affect everyone.