This is a personal note.
Think about the copyright of game images.
→The copyright is held by the company that created the game, but YouTube videos of gameplay are acceptable because they can be used to advertise the game.
On the other hand, altering or making memory cards with illegal save data is copyright infringement.
From this
I wonder if a similar scheme can be considered for automatically generated AI products.
After all, OK game videos are also a kind of co-regulatory approach.
If this is the case, AI companies would have a certain amount of copyright protection (depending on the product), but they would also have to set a borderline for the division of rights and limits with users (in a co-regulatory manner, an AI organization or something would set standards in a timely and precise manner according to the technical level). I wonder if it is possible to think of such a range of rights extension that users can use freely within that range.
Of course, the theory of responsibility is also inextricably linked.
I think this is a research issue.
