But unlike Google’s version, Claude can accidentally regurgitate the entire text or passages from it, yes?
So it’s not really internal and this judge is an imbecile, correct?
(I know that previous “AI” engines have been tricked into returning the original paintings and faces of people that they had ingested, so I assume this is also a possibility for this “AI” too.)
Side note: I used to backtrace Midjourney’s “art” to the original non-public domain images they came from.
Weird queries would lead to the same faces created every time, and if you’ve ever played semantle you could find the original art by doing a hotter/colder process.
Still don’t know how they haven’t been shut down for copyright infringement.
But unlike Google’s version, Claude can accidentally regurgitate the entire text or passages from it, yes?
So it’s not really internal and this judge is an imbecile, correct?
(I know that previous “AI” engines have been tricked into returning the original paintings and faces of people that they had ingested, so I assume this is also a possibility for this “AI” too.)
Side note: I used to backtrace Midjourney’s “art” to the original non-public domain images they came from.
Weird queries would lead to the same faces created every time, and if you’ve ever played semantle you could find the original art by doing a hotter/colder process.
Still don’t know how they haven’t been shut down for copyright infringement.
According to the information provided to the judge, including as claimed by the plaintiffs, no. Their core complaint is only the training.