Hacker News Reader

> > Isn't OpenAI mostly built upon disregarding the copyright of countless people?

> It sure was.

Can you cite something that elaborates on this point? Do people who read books and then learn from it also disregard copyright? How is what OpenAI does meaningfully different from what people do?

View on HN

Are those people then reselling the contents of the books?

How is that relevant? Is OpenAI reselling the contests of books? If so, this is the first I have heard of it.

Yes. You can (could?) paste the “article preview” into GPT-4 and get full articles for free, basically pirating a NYT subscription. Here are “one hundred examples of GPT-4 memorizing content from the New York Times”:

https://nytco-assets.nytimes.com/2023/12/Lawsuit-Document-dk...

Thanks for sharing, that is indeed quite a compelling case, and I did not expect it to be regurgitating verbatim text to this extent. It would be interesting to see how OpenAI addresses this.

Thanks for being open to adjusting your opinion when given new information!

DALL-E replicates artist’s lifestyle with no regard to copyright.

Sora is doing the same on YouTube videos. It blocks queries like “in the style of Wes Anderson” but it still uses that as training data and generating content.

Reply on news.ycombinator.com