David Colarusso
If you only read one article on the NYT suing OpenAI, make it this one from @mmasnick
This isn't about copyright in the traditional sense. At best it's a negotiating tactic to get better terms in an agreement with OpenAI, at worst it's a push to create a bunch of "rights" out of thin air. Copyright is not the way to deal with "AI!"
Paul Cantrell
@Colarusso @mmasnick
Thanks for this much needed context.
I wonder if the piece is a bit too dismissive of the part about the near-literal reproduction of training data. LLMs do things that would certainly be considered plagiarism if done by a human, and not only when prompted with large chunks of the source text. In my legal ignorance, that does seem relevant to copyright, at least potentially infringing.
(The central point about copyright not prohibiting reading stands regardless.)
Paul Cantrell
@Colarusso @mmasnick Understood and agreed.
The submitted evidence for copying in this NYT case seems rather weak to me; they have (as Masnick points out) set up highly concocted circumstances to get that output, but GPT will also repeat near-verbatim chunks under much looser prompting.
I’m thinking in particular of legal risks for companies using this tech, who could plausibly find themselves sued on copyright grounds for using GPT output.
David Colarusso
@inthehands I agree that a focus on output is the strongest part of the NYT’s argument (and what’s been missing from most of these cases), but there is something to be said for countering the impression that these systems just spit out copies of others’ work without a lot of effort. One has to work to get those nearly identical outputs, otherwise you’re left with paraphrased content, which isn’t a copy. @mmasnick
Paul Cantrell
@Colarusso @mmasnick This, for example, seems to me (again, in my ignorance) to flirt with the legal boundaries of copyright infringement: https://futurism.com/cnet-ai-plagiarism I’m not sure that infringement claim would make it through court, or be worth the payoff if it it, but the legal risk is greater than zero.
(Again, agreed with larger points about copyright not being the best or most relevant legal regime here, just not completely irrelevant.)
Paul Cantrell
@Colarusso @mmasnick Indeed. In moments of optimism, I wonder how much sanity we could achieve with nothing more than legal clarity that companies are in fact liable for decisions under their control, regardless of whether they delegated those decisions to an employee, a third party, or a machine. Then insurance-like frameworks like Copyright Shield take on an actuarial (as opposed to PR) character, and…better outcomes, maybe??
David Colarusso
@inthehands by no means is it irrelevant. There is a reason OpenAI introduced Copyright Shield. https://techcrunch.com/2023/11/06/openai-promises-to-defend-business-customers-against-copyright-claims/amp/ @mmasnick
GNU social JP is a social network, courtesy of GNU social JP管理人. It runs on GNU social, version 2.0.2-dev, available under the GNU Affero General Public License.
All GNU social JP content and data are available under the Creative Commons Attribution 3.0 license.