Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

NY Times copyright suit wants OpenAI to delete all GPT instances

arstechnica.com

Post reply on HN
Print view View original

521–530 of 921 posts

Re: NY Times copyright suit wants OpenAI to delete all GPT instances

#521

Post by Matticus_Rex »

eigenket wrote: post #494
Matticus_Rex wrote: post #487

Earlier quoted context omitted.

It turns out you can reproduce articles with next-token prediction when the articles are quoted all over the dataset. The articles themselves are indisputably not a part of the model, because it doesn't store text at all. OpenAI's position is correct; people just underestimated how well the AI learns from reading, especially when it reads the same text in a bunch of different places because it's being quoted/excerpte…

If it can and does reproduce a piece of text verbatim then the text is indisputably stored somehow in the model.

That's just not true. There's no search and retrieval involved. It just associates the words so strongly in that context because they were in the training data so often that next-token prediction can (sometimes, in some limited circumstances) reproduce chunks of it. It's like if a human had read pieces of an article so many times and knew NYT style so well that they could spit out chunks of an article verbatim, but using more efficient hardware and with no actual self-understanding of what it's doing.

Re: NY Times copyright suit wants OpenAI to delete all GPT instances

#522

Post by sideshowb »

groceryheist wrote: post #90
visarga wrote: post #82

Earlier quoted context omitted.

> Just learn to recognize and punish plagiarism via RLHF. This is not a RLHF problem. What I was expecting them to do is to keep a bloom filter of ngrams for known copyrighted content, such as enumerating all sets of n=7 consecutive words in an article, and validate against it. The model would only output at maximum n-1 words that look verbatim from the source. But this will blow up in their face. Let's see: - AI com…

Maybe the bloom filter solution is enough, but I wonder. - Paraphrasing n=7 words (and quite a few more) within a sentence can easily be fair use. - As n gets big, the bloom filter has to also. If/when attribution is solved for LLMs (and not fake attribution like from Bing or Perplexity) then creators can be compensated when their works are used in AI outputs. If compensation is high enough this can greatly incentivi…

I like the idea but seems like there would be big problems. Like detecting if a work is reworded. Or a large number of sources have all slightly influenced a small response - isn't that pretty much considered new knowledge?

Then there's the issue that however you credit attribution, it creates a game of enshittified content creation with the aim of being attributed as often as possible, regardless of whether the content really offered anything that wasn't out there already.

Re: NY Times copyright suit wants OpenAI to delete all GPT instances

#523

Post by thfuran »

tzs wrote: post #363
munchinator wrote: post #180

It's interesting to me the ambiguous attitude people have to reproducing news content. Whenever there is a story from NYT on HN (or any other large media outlet), the top comment is almost always a link to an archived version which reproduces the text verbatim. And this seems to be tolerated as the norm. And yet, whenever there is a submission about a book, a TV show, a movie, a video game, an album, a comic book, or…

> I think that's something worth reflecting on, about why we feel it's OK to pirate news articles, but not other IP As you noted it is not the norm to post pirate links here for IP other than news articles, but that doesn't mean that a lot of people think it is not OK to pirate those other forms of IP. In nearly any big discussion that even remotely involves video streaming there will be numerous posts from people ex…

>"subscribing is not an option because even though this paid service does exactly what I want now at a price that is trivial for me they might someday later change"

I'm not saying you've never seen anyone make an argument roughly like that, but I will certainly say that it is not at all representative of the argument that I see made. Complaints usually have to do with current behavior of the platform or the wider streaming ecosystem.

Re: NY Times copyright suit wants OpenAI to delete all GPT instances

#524

Post by djhope99 »

This argument that the LLM is learning seems slightly flawed when you consider that other experts in the field consider it more like lossy compression. If it’s lossy compression that’s really happening here then you can understand the copyright argument. It’ll be interesting to see how this plays out, lots of new ground breaking.

Re: NY Times copyright suit wants OpenAI to delete all GPT instances

#526

Post by phpisthebest »

gnz11 wrote: post #471
phpisthebest wrote: post #420

Earlier quoted context omitted.

None of which are copyrightable and infact has been the subject of DMCA abuse like when a Movie uses NASA footage and claims copyright on YouTube videos with the same footage. Copyright is a complex subject, and not as vast as many believe, at the same time ironically it is more vast than i believe it should be. copyright should be much more limiting than it is. Which is at odds with people that believe copyright sho…

Copyright doesn’t exist solely for the “promotion of useful sciences”. https://en.m.wikipedia.org/wiki/Copyright

Citing Wikipedia you already failed..

That is a General Article about Copyright world wide, I Specifically stated US Copyright, which is Authorized by Article I, Section 8, Clause 8 of the United States Constitution[1], implicitly for the promotion of the useful sciences. That is where congress derives its power to pass copyright laws, and to enforce copyright on the people of the United States. No other purpose is authorized by the US Constitution

[1] https://www.law.cornell.edu/wex/intellectual_property_clause

Re: NY Times copyright suit wants OpenAI to delete all GPT instances

#528

Post by Robotbeat »

heavyset_go wrote: post #311
pas wrote: post #195

Earlier quoted context omitted.

sure, but if I use an LLM to write a novel/article, I can be sued in civil court not the LLM. but, more importantly, OpenAI can also be sued for tortious interference? (basically the civil equivalent of accessory)

Whoever operates the LLM, in this case OpenAI, engaged in copyright infringement through the unauthorized modification, reproduction and distribution of content to you.

The person doing the requesting did.

Re: NY Times copyright suit wants OpenAI to delete all GPT instances

#529

Post by jasoneckert »

munchinator wrote: post #180

It's interesting to me the ambiguous attitude people have to reproducing news content. Whenever there is a story from NYT on HN (or any other large media outlet), the top comment is almost always a link to an archived version which reproduces the text verbatim. And this seems to be tolerated as the norm. And yet, whenever there is a submission about a book, a TV show, a movie, a video game, an album, a comic book, or…

I believe the reason many of us tolerate links to news articles and other content is because we believe in equality when it comes to information access. In other words, many of us believe that those who cannot afford a subscription to a paywalled site should still be able to read the articles, in much the same way public libraries allow those who cannot afford to purchase a book the ability to read it.

However, this doesn't apply to organizations that freely share copyrighted information while making money in the process, or to organizations that share copyrighted information in a way that specifically disadvantages or does harm to the original creator of that information.

Post reply on HN