Hi,
When using chatgpt to translate my own text from one language to another, do I own the full copyrights of the translated text?
I guess that means that the translation result is considered as a derivative work, still owned by me.
By language I mean natural language, not, e.g., a programming language.
This is not legal advice!!
Its a super gray area. Translations need to be seen as “subjective artistic” work to be fully protected i beleive, so the method will probably matter.
I think if you’re translating your own work, using GPT for the task doesnt attribute any ownership to to OpenAI, however you should 100% disclose that it was done with AI.
Conversly if you’re tranlating someone elses work, id argue that using AI means you havent done any personal subjective “artistic” translation and would most likely infringe on copyright if you try to claim authorship in any way. At that point it feels the same as paying a service.
Isn’t this the same as compiling program code into executable? It is quite clear that the owner of the code owns the executable as well, even if no “artistic effort” is involved in compilation process.
Yes? I think we’re saying the same thing. Using the tool to translate your own content does nothing to change your authorship rights, and the translation is fully yours as well…but you MUST disclose the use of AI.
Using the tool to translate someone elses work , in my opinion, removes all artistic or personal interpretive component from the process, and claiming ownership would infringe on the authors rights.
Yes, we do agree on these 
However, what do you mean by “must disclose use of AI”. To who, and why?
Say, if I were an author of a book, and then translate it to another language using chatgpt. Perhaps I want to offer it to a foreign publisher. Why would that translation information need to be disclosed to anyone?
Considering the programming analogue; it would be the same as a SW company would need to disclose technical specs of the compiler to customers (and probably many other tools in the dev pipeline), which is ridiculous idea.
[quote=“perza, post:5, topic:293854”] it
would be the same as a SW company would need to disclose technical specs of the compiler to customers (and probably many other tools in the dev pipeline), which is ridiculous idea.
[/quote]
Okay I see what you mean, and no that comparison doesn’t line up because compiling and translating are completely different.
If the foreign publisher decides they don’t want to pay you, they can try to claim you don’t have rights to the translation because you presented it as your own full work, when you had a “co-author” for the translation.
Just look at some of the recent AI copyright cases. Every time they strongly suggest disclosing the use of AI to avoid all this
Firstly, I see no problem providing the information of the translation tool in the case mentioned.
Secondly, thanks for the information about the copyright cases. It is insane that this kind of claims are made, but it is good to know.
Fundamentally, current AIs like chatgpt are computer algorithms. They are not conscious or creative or anything like that . Thus a code compiler and language translator are identical in their function, even if the latter is vastly more complex.