Anthropic pirating books for their training corpus resulted in the biggest copyright settlement in history–well over a billion. That is still being quibbled over i believe, but they settled because they were likely to pay out more if the case went forward. So I’m not really sure where you’re coming from that infringement via torrenting does not result in monstrously large liability.
This is wildly wrong in so many ways.
Copyright is an intellectual property right, firmly grounded in property law doctrine–you are probably thinking of trademark, which is rooted in consumer protection law, or likeness rights which have their roots in privacy law.
The copyrightability of AI generated content gets to where the nexus of creativity happens. Effectively, image generators (modern ones–i actually don’t think DABUS is a diffusion model) are operated like a commissioned work. The user gives detailed instruction on par with what you might see in a commissioned work, and the creative event occurs when the “contractor” interprets that into the work. The copyright may be assigned or it may be licensed, in any case, the initial copyright holder is the contractor–or in our case, the model. Now, it is well established that only humans can have sua sponte property rights, including intellectual property right. Those can be assigned, licensed, etc., but they must first inher to a human and so an AI system literally has no copyright to assign, were it even able to engage in a contractual agreement to transfer said rights. As a result, no, there is no copyright in AI generated content and without a significant change in law there is unlikely to ever be any.
If he had sought to register the copyright under purely his own name, he would have been committing a fraud on the copyright office. This wasn’t explicitly established at the time of his suit, but it has been very explicitly the case now for over a year. When registering copyright you must declare any AI-generated components. Failure, or refusal, to do so constitutes a fraud on the office and such fraud is sanctionable up to revoking the copyright in the work in its entirety, even if the AI-gen component was only partial. This is really important to note with software copyright and the kind of litigation we’re likely to see wrt piracy in the future (i.e., defendant claims plaintiffs did not declare vibe coded components and thus committed a fraud on the office and should be sanctioned with full revocation of the right as a signal to other would-be claimants).
Just noticed your reply and want to correct this. Anyhropic settled, the 1.5bil was not a judgment against them. Specifically, this covered the literal pirating of the training corpus. It had absolutely nothing to do with the way training on the data handled the training data–they literally torrented an enormous portion of their training corpus.
Anthropoc DID try to argue that because they used the pirated material for training a model, it was fair use. The judge correctly decided that doesn’t make any fucking sense. Again, this is not about the models encoding data, it is literally just about the fact that these silly fucks torrented vast portions of their training corpus like college students building a porn library on college broadband.