Comment on OpenAI claims The New York Times tricked ChatGPT into copying its articles

<- View Parent
gmtom@lemmy.world ⁨10⁩ ⁨months⁩ ago

it still manages to write Harry Potter. If a human did the same thing, even if they honestly and genuinely thought they were presenting original ideas, they would still be guilty.

Only if they publish or sell it. Which is why OpenAI isnt/shouldn’t be liable in this case.

If you write out the entire Harry Potter series from memory, you are not breaking any laws just by doing so. Same as if you use photoshop to reproduce a copyright work.

So because they publish the tool, not the actual content openAI isn’t breaking any laws either. It’s much the same way that torrent engines are legal despite what they are used for.

There is also some more direct president for this. There is a website called “library of babel” that has used some clever maths to publish every combination of characters up to 3260 characters long. Which contains, by definition, anything below that limit that is copywritten, and in theory you could piece together the entire Harry Potter series from that website 3k characters at a time. And that is safe under copywrite law.

The same with making a program that generates digital pictures where all the pixels are set randomly. That program, if given enough time /luck will be capable of generating any copyright image, can generate photos of sensitive documents or nudes of celebrities, but is also protected by copyright law, regardless of how closely the products match the copyright material. If the person using the program publishes those pictures, that a different story, much like someone publishing a NYT article generated by GPT would be liable.

source
Sort:hotnewtop