And suddenly the Internet is gung-ho in favor of EULAs being enforceable simply by reading the content the website has already provided.
Recent major court cases have held that the training of an AI model is fair use and doesn’t involve copyright violation, so I don’t think licensing actually matters in this case. They’d have to put the content behind a paywall to stop the trainer from seeing it in the first place.
Nope, this was one of them. The case had two parts, one about the training and one about the downloading of pirated books. The judge issued a preliminary judgment about the training part, that was declared fair use without any further need to address it in trial. The downloading was what was proceeding to trial and what the settlement offer was about.
Doesn’t really matter what I think, its a different concept than pirating. Hence a different thing than what was getting ruled on.
I mean AI or not look at it this way: if a company wanted to train their workers and pirated all the training manuals, piracy is the issue, not the training.
Is it hypocrisy to be for EULA enforcement on reading when it’s machines, but not when it’s humans? Crawlers “read” on a massive scale that doesn’t compare to humans.
I don’t think so, or not always. humans need to find the EULA on the website by first loading the main page or another they found a link to. but if the path of that document was standardized, it could be enforced that way for robots
And suddenly the Internet is gung-ho in favor of EULAs being enforceable simply by reading the content the website has already provided.
Recent major court cases have held that the training of an AI model is fair use and doesn’t involve copyright violation, so I don’t think licensing actually matters in this case. They’d have to put the content behind a paywall to stop the trainer from seeing it in the first place.
I guess that’s a different court case than the one where Anthropic offered to pay $1.5 billion?
Nope, this was one of them. The case had two parts, one about the training and one about the downloading of pirated books. The judge issued a preliminary judgment about the training part, that was declared fair use without any further need to address it in trial. The downloading was what was proceeding to trial and what the settlement offer was about.
Totally different. Anthropic could have bought all the books and trained on them. Pirating is a different topic.
You think buying the books would let them plagiarize ? That doesn’t seem to be normal in the “book buying” process.
Doesn’t really matter what I think, its a different concept than pirating. Hence a different thing than what was getting ruled on.
I mean AI or not look at it this way: if a company wanted to train their workers and pirated all the training manuals, piracy is the issue, not the training.
Given the judege in that case flat out rejected the claim that there was any infringement for works they had legally aquired, yes.
Is it hypocrisy to be for EULA enforcement on reading when it’s machines, but not when it’s humans? Crawlers “read” on a massive scale that doesn’t compare to humans.
I don’t think so, or not always. humans need to find the EULA on the website by first loading the main page or another they found a link to. but if the path of that document was standardized, it could be enforced that way for robots