5 ms·

You do copyright for content that you invented and which didn't exist before. But NYT content is reporting on events truthfully to the public without any ficti

You do copyright for content that you invented and which didn't exist before. But NYT content is reporting on events truthfully to the public without any fiction or lies. Since there can be only one truth it should not matter whether NYT or Washington Post or ChatGPT is spinning it out. Unless NYT is claiming they don't report truth and publishes fiction. That is of concern since, NYT claims to reporth news truthfully. So is NYT scamming Americans hundreds of millions of dollars by charging for subscription fees by making a false promise on things that they report? This should be the bigger question here.

artninja1988 3y ago
AFAIK facts like happenings in the world are not copyrightable. So I guess the nyt is arguing it's copying their prose and way of writing about them?
- mrkeen 3y ago
  Yes. Journalism is a job. People do the work of turning these happenings into words, and are paid for it. That's what's stolen here. The value created through doing that work. If it didn't have value, Microsoft would lose nothing by no longer ingesting it.
  - rand1239 3y ago
    > People do the work of turning these happenings into words, and are paid for it. That's what's stolen here. Stolen from whom? Journalists who got reported got paid. The owner is a billionaire. I don't understand your logic. Does NYT pays money to the people/countries etc it uses to as subject to create content(NEWS)? Isn't that stealing then? Also their website TOS didn't prohibit LLMs from using their data.
    - mrkeen 3y ago
      > Stolen from whom? The owner is a billionaire. > ...owner... > Does NYT pays money to the people/countries etc it uses to as subject to create content(NEWS)? Isn't that stealing then? No, that's why in my reply to "facts like happenings in the world are not copyrightable" I emphasised do the work. Journalism is a job. Happenings do not just fall onto the page. > Also their website TOS didn't prohibit LLMs from using their data. This is just lazy. We have rule of law. Individuals don't need to write "don't break law X" to be protected by them. And nytimes does in fact have copyright symbols on its pages - not that it needs them.
      - rand1239 3y ago
        There is no rule of law saying LLMs cannot be trained on WWW data. New York times made it ridiculously easy for anyone to access their content by putting it in WWW for making money from page impressions. And they started ingesting links of their content to social media, search engines, etc. And now they are acting surprised someone used the content to train an LLM. Should have done their job in the first place to prevent it from training LLMs and make it less. But they didn't because that affects their page impressions and ad views. Because the more open the content the more money they make everyone click on a link and see the ad. You can't have it both ways. If you do gambling by making content so open so you can get more views from ads, you also get to enjoy the consequences and not cry like a baby asking for billions by making stupid decisions in the first place.
lionkor 3y ago
> You do copyright for content that you invented and which didn't exist before. I dont think that's accurate. The Copyright Act, § 103, allows copyright protection for "compilations (of facts)", as long as there is some "creative" or "original" act involved in developing the compilation, such as in the selection (deciding which facts to include or exclude) and arrangement (how facts are displayed and in what order).
- rand1239 3y ago
  Okay. But ChatGPT doesn't spin out the fact in the same order right? So how does this stand in court?
  - lionkor 3y ago
    as far as I understodd ChatGPT reproduced a word-for-word part of an NYT article(?), but not sure, didnt read the full post yet.
passwordoops 3y ago
Not sure where you're coming from in this. A NYT article, once written is copyrighted. Using the content without attribution is at best plagiarism, and spitting it out the way the LLMs do is definitely a violation of if that copyright. Unless you're telling me ChatGPT has eyes and sources just like the NYT and is worrying events as it sees them too?
- rand1239 3y ago
  I don't understand. So if New York times reported on a new laws of physics and put as an article will became copyrighted? Nobody would be able to talk about it and has to discover it by themselves? How is reporting on an event different from reporting on discovering a scientific law?
  - passwordoops 3y ago
    It's not. That's why at the end of every article that is not original reporting you will find a little bit saying "As originally reported by (organization)" and there is usually some sort of license associated with that. ChatGPT neither includes sources nor deals with any licensing. That's the issue
  - jprete 3y ago
    The exact words used to explain the scientific law are copyrighted by the writer (presumably the paper's authors). Rephrasings are not copywrited by the source, but by the rephrasing entity (e.g. the NYT, or a teacher that made a handout for their class). Copyright on scientific papers is most definitely a thing, by the way.
    - rand1239 3y ago
      If the bar for copyright is as low as ordering of words, then I don't even know what to say.
      - passwordoops 3y ago
        So stop saying anything. Go learn how copyright works in the real world
noitpmeder 3y ago
Is your position that all non fiction textual work is uncopywritable?
- rand1239 3y ago
  Yeah. I don't think it makes much sense to allow copyright on textual descriptions of events that happened. Now the question is whether did OpenAI violate the terms of service by using the bits transferred from NYT to train their LLM. I don't think their TOS had LLMs mentioned. So it's on NYT to be negligent and not update their TOS right?

artninja1988 3y ago

AFAIK facts like happenings in the world are not copyrightable. So I guess the nyt is arguing it's copying their prose and way of writing about them?

mrkeen 3y ago
Yes. Journalism is a job. People do the work of turning these happenings into words, and are paid for it. That's what's stolen here. The value created through doing that work. If it didn't have value, Microsoft would lose nothing by no longer ingesting it.
- rand1239 3y ago
  > People do the work of turning these happenings into words, and are paid for it. That's what's stolen here. Stolen from whom? Journalists who got reported got paid. The owner is a billionaire. I don't understand your logic. Does NYT pays money to the people/countries etc it uses to as subject to create content(NEWS)? Isn't that stealing then? Also their website TOS didn't prohibit LLMs from using their data.
  - mrkeen 3y ago
    > Stolen from whom? The owner is a billionaire. > ...owner... > Does NYT pays money to the people/countries etc it uses to as subject to create content(NEWS)? Isn't that stealing then? No, that's why in my reply to "facts like happenings in the world are not copyrightable" I emphasised do the work. Journalism is a job. Happenings do not just fall onto the page. > Also their website TOS didn't prohibit LLMs from using their data. This is just lazy. We have rule of law. Individuals don't need to write "don't break law X" to be protected by them. And nytimes does in fact have copyright symbols on its pages - not that it needs them.
    - rand1239 3y ago
      There is no rule of law saying LLMs cannot be trained on WWW data. New York times made it ridiculously easy for anyone to access their content by putting it in WWW for making money from page impressions. And they started ingesting links of their content to social media, search engines, etc. And now they are acting surprised someone used the content to train an LLM. Should have done their job in the first place to prevent it from training LLMs and make it less. But they didn't because that affects their page impressions and ad views. Because the more open the content the more money they make everyone click on a link and see the ad. You can't have it both ways. If you do gambling by making content so open so you can get more views from ads, you also get to enjoy the consequences and not cry like a baby asking for billions by making stupid decisions in the first place.

lionkor 3y ago

> You do copyright for content that you invented and which didn't exist before. I dont think that's accurate. The Copyright Act, § 103, allows copyright protection for "compilations (of facts)", as long as there is some "creative" or "original" act involved in developing the compilation, such as in the selection (deciding which facts to include or exclude) and arrangement (how facts are displayed and in what order).

rand1239 3y ago
Okay. But ChatGPT doesn't spin out the fact in the same order right? So how does this stand in court?
- lionkor 3y ago
  as far as I understodd ChatGPT reproduced a word-for-word part of an NYT article(?), but not sure, didnt read the full post yet.

passwordoops 3y ago

Not sure where you're coming from in this. A NYT article, once written is copyrighted. Using the content without attribution is at best plagiarism, and spitting it out the way the LLMs do is definitely a violation of if that copyright. Unless you're telling me ChatGPT has eyes and sources just like the NYT and is worrying events as it sees them too?

rand1239 3y ago
I don't understand. So if New York times reported on a new laws of physics and put as an article will became copyrighted? Nobody would be able to talk about it and has to discover it by themselves? How is reporting on an event different from reporting on discovering a scientific law?
- passwordoops 3y ago
  It's not. That's why at the end of every article that is not original reporting you will find a little bit saying "As originally reported by (organization)" and there is usually some sort of license associated with that. ChatGPT neither includes sources nor deals with any licensing. That's the issue
- jprete 3y ago
  The exact words used to explain the scientific law are copyrighted by the writer (presumably the paper's authors). Rephrasings are not copywrited by the source, but by the rephrasing entity (e.g. the NYT, or a teacher that made a handout for their class). Copyright on scientific papers is most definitely a thing, by the way.
  - rand1239 3y ago
    If the bar for copyright is as low as ordering of words, then I don't even know what to say.
    - passwordoops 3y ago
      So stop saying anything. Go learn how copyright works in the real world

noitpmeder 3y ago

Is your position that all non fiction textual work is uncopywritable?

rand1239 3y ago
Yeah. I don't think it makes much sense to allow copyright on textual descriptions of events that happened. Now the question is whether did OpenAI violate the terms of service by using the bits transferred from NYT to train their LLM. I don't think their TOS had LLMs mentioned. So it's on NYT to be negligent and not update their TOS right?