Copyright works on the premise that creating a new creative work is hard and requires a considerable investment. When machines start creating translations, the investment becomes far smaller, and as a result, a copyright on such "works" has little value, because everybody who requires a translation can ask the AI to produce yet another translation (and with improved AI's, the quality will probably be better anyway, although I think it will take a long time before AI can replace humans with regard to literary works). -- here I am disregarding the initial investment that goes into building the AI itself. (Which is now a complete bubble, in which a mere million is considered pocket change.) Of course, the current promoters of AI want to both have their cake and eat it, and may at the same time try to claim copyright on their machines output, and deny rewarding the value extracted from our collective public (and not so public) cultural domain.... I don't know yet how the law will work out in the long term, but for now, pure AI output is public domain ab initio, unless it can be shown to be derived from actual copyrighted works (in which case it is most likely copyright infringement). Jeroen. James Adcock via gutvol-d schreef op 2026-01-07 02:47:
Not sure to what extent current language translation should be considered "AI", but one can get a sense of the coming issues by searching on "copyright of materials generated by ai".
In USA, but not necessarily in other countries, current bias is against giving copyright where AI was the primary form of generation — but not if it is only used as a "tool" as part of a larger human-creative process.
I think one might be concerned of a coming mess of translations and other "AI" generation based on "risen to public domain" works — and of human beings claiming copyright on parts of that mess.
Jim Adcock