Earlier quoted context omitted.
If I read a story about a flood in Dubai on the NYTimes and then I write an email to my friend summarizing what I just read, it is not copyright infringement. I am not sure why it would suddenly become infringement because an LLM is composing that email for me.
But if you create a company that takes the NYT articles, scans the articles for extreme weather information in Dubai and emails your paying customers every time that happens, then you are doing copyright infringement.
New York Times considers legal action against OpenAI as copyright tensions swirl
371–380 of 383 posts
Re: New York Times considers legal action against OpenAI as copyright tensions swirl
#372Earlier quoted context omitted.
Let's assume that happens. How do I hedge against it? Is there a convenient way to mirror the bits of the web that are open now? Perhaps a mirror of archive/WayBack machine that could be viewed locally similar to Wikipedia dumps?
Yes, commoncrawl indexes
Re: New York Times considers legal action against OpenAI as copyright tensions swirl
#373Earlier quoted context omitted.
Search engines provide links, but also titles and snippets of the page -- enough for you to decide if you want to visit, and Google will show you their cached page if you ask for it. Even the link is a copyrightable item -- artistic effort went into creating it
> Search engines provide links, but also titles and snippets of the page -- enough for you to decide if you want to visit Small snippets are allowed by copyright law. They are not infringing. > and Google will show you their cached page if you ask for it. Really? I haven't seen that in several years. How do you get it these days? I always assumed Google quit giving you that option exactly because of copyright issues.…
http://webcache.googleusercontent.com/search?q=cache:www.hac...
etc
There's also a link in the three-dot menu next to the search result, but it doesn't always appear.
Re: New York Times considers legal action against OpenAI as copyright tensions swirl
#374Re: New York Times considers legal action against OpenAI as copyright tensions swirl
#375Earlier quoted context omitted.
Yes, commoncrawl indexes
100TB. Dang! I will need to expand my storage significantly to mirror it.
And that's probably just 1 year. There's data that's vanished over time. The point of all this is that they're point-in-time references of the internet.
the more practical thing is that (the royal) they have set up a rclone like utility that allows you to retrieve a slice of the whole snapshot, if you know your domain.
it would be extremely funni if commoncrawl was a de-facto ipfs, allowing queries along domains (or arbitrary dimensions/clustering) of interest.
Re: New York Times considers legal action against OpenAI as copyright tensions swirl
#376Earlier quoted context omitted.
> First, a human didn’t create it, so copyright protections don’t exist on its output. Machines don’t enjoy copyright protections, people do. This is not correct. AI models are tools that humans use. This is like saying "it was typed on a computer therefore it doesn't enjoy copyright protections"
This has a ruling, see https://www.theartnewspaper.com/2023/05/04/us-copyright-offi... When you're using ai models to generate the art, it's not considered human enough. If you then make a bunch of modifications to it, sure, but giving the initial prompt is currently insufficient
Interesting and makes sense. I guess the best comparison I can think of is a random number generated with Math.random() is not copyrightable. But including that random number is a program is copyrightable.
I'd say that a corollary to this is that if the images/words generated by the machine are not copyrightable then they cannot also be _copyright violations_
Re: New York Times considers legal action against OpenAI as copyright tensions swirl
#377Earlier quoted context omitted.
A fantasy novel with Tolkien's characters' names is an evident copyright violation regardless of how it was generated. That's not what's happening here.
No, but the point is that OpenAI expects to be held harmless if someone uses its tools to violate trademarks [unlike if a human employee had been paid to read Tolkein and write a story about Gandalf and hobbits] because its not an OpenAI employee consciously doing it, but just a model that translates its given inputs If GPT is blameless doing some things because it's a deterministic model, not an agent, then the "it…
Re: New York Times considers legal action against OpenAI as copyright tensions swirl
#378Earlier quoted context omitted.
> How would one even go about a phonebook-style mechanical listing of facts and occurrences? The traditional trick there is to include some small amount of fake data in the directory. You know someone has copied your collection of facts instead of compiling their own because it includes your fake facts. Mapmakers have used the method for at least as long as cartography has been part of our recorded history, see https…
How are trap streets “the trick” if they are not copyrightable or enforceable under US law? Like the Supreme Court has specifically considered and dismissed this “one clever trick to keep people from copying lists of facts” because to allow it would completely undermine the idea of facts and figures not being copyrightable. > Trap streets are not copyrightable under the federal law of the United States. In Nester's M…
US law is not the only law.
Current US law has not been as it is for the entire existence of the US.
Trap streets and other such devices have existed much longer than the US.
As well as copyright law the trick can help detect beaches in contractual agreements that cover use of information from services. Action based on such breaches do not necessarily end up in a court of law at all.
Out of court settlements do not necessarily rely on the letter (nor intent) of the law, but often instead the expense (money directly, time, potential reputational risk) of defending a position even if the law is on your side. The threat of action is often enough to make the other party cave and such actions will usually happen well out of public view (I know of one instance involving a list of phone numbers, that I won't go into in detail because while there is no NDA or such in force this discussion is not worth irritating people I have the confidence of!).
> clumsy attempt to work around this.
That the trick is clumsy does not mean it isn't still commonly used (it absolutely is) or that it has not been used in successful cases between map makers and such (it has, one significant example is given in the paragraph directly following the one you selected to quote from the page linked in my previous post).
Re: New York Times considers legal action against OpenAI as copyright tensions swirl
#379Earlier quoted context omitted.
> my opinion is that if your AI uses GPL code the output should be GPL If I use Inkscape is the output of my drawing subject to the same terms as Inkscape? If I use a Photoshop filter is the output subject to Photoshop's EULA and/or the copyright of the photo I started with?
If I use a Photoshop filter is the output subject to Photoshop's EULA and/or the copyright of the photo I started with? If you get my image from the internet then you resize it in Photoshop you can't claim you created some original art, you just used the resize/crop/color filter function.
Re: New York Times considers legal action against OpenAI as copyright tensions swirl
#380Earlier quoted context omitted.
If I use a Photoshop filter is the output subject to Photoshop's EULA and/or the copyright of the photo I started with? If you get my image from the internet then you resize it in Photoshop you can't claim you created some original art, you just used the resize/crop/color filter function.
Indeed. I was hoping to draw the distinction between license of the tool vs its output. AI is interesting in that its training inputs can leak out, sometimes in ways that are verbatim or insufficiently transformative to be free of copyright.
What happens is that Microsoft or OpenAI will use 1000 or more different sources and then the author is not clear unless in few outputs where the tool outputs the exact code as the training data. If the AI is actual inteligent they could train it in a few books and say MDN documentation and not a few thousands of GPL code.