Hacker News Reader

Comment by oh_sigh | Hacker News Reader

Does the law care about the method(whether it is ML or just reading the article from a database) if the model can spit out a verbatim copy of an article?

View on HN

I see this argument everywhere and have yet to succeed in getting ChatGPT to give an article. This argument is basically a non-starter

Researchers have gotten large chunks of verbatim text. Suggesting it’s only infringing if the entire work is reproduced is legally incorrect and therefore irrelevant.

https://mashable.com/article/chatgpt-revealed-personal-data-...

> Suggesting it’s only infringing if the entire work is reproduced is legally incorrect

It may not be conclusive, but it's highly suggestive:

> Amount and substantiality of the portion used in relation to the copyrighted work as a whole: Under this factor, courts look at both the quantity and quality of the copyrighted material that was used. If the use includes a large portion of the copyrighted work, fair use is less likely to be found; if the use employs only a small amount of copyrighted material, fair use is more likely.

> https://www.copyright.gov/fair-use/

Context is important when reading that, just 7 words can be copyright infringement.

I've seen many of these "AI" offerings regurgitate their training data verbatim. If chatgpt were miraculously immune to this, despite taking the same basic approach as everybody else, and without significant efforts to prevent regurgitation, then the onus is on them to prove that.

It's not an argument, it is a question. I'd like to know the answer as to whether the method matters to the courts.

Reply on news.ycombinator.com