zlacker

[parent] [thread] 2 comments
1. safety+(OP)[view] [source] 2023-12-27 15:14:22
The NYT's strongest argument for infringement is that OpenAI is reproducing their content verbatim (and to make matters worse, without attribution). IANAL but it seems super likely to me that this will be found to be infringing sooner or later.

Do I really want to use a Chinese word processor that spits unattributed passages from the NYT into the articles I write? Once I publish that to my blog now I'm infringing and I can get sued too. Point is I don't see how output which complies with copyright law makes an LLM inferior.

The argument applies equally to code, if your use of ChatGPT, OpenAI etc. today is extensive enough, who knows what copyrighted material you may have incorporated illegally into your codebase? Ignorance is not a legal defense for infringement.

If anything it's a competitive advantage if someone develops a model which I can use without fear of infringement.

Edit: To me this all parallels Uber and AirBnB in a big way. OpenAI is just another big tech company that knew they were going to break the law on a massive scale, and said look this is disruptive and we want to be first to market, so we'll just do it and litigate the consequences. I don't think the situation is that exotic. Being giant lawbreakers has not put Uber or AirBnB out of business yet.

replies(1): >>jterry+F5
2. jterry+F5[view] [source] 2023-12-27 15:46:34
>>safety+(OP)
>IANAL but it seems super likely to me that this will be found to be infringing sooner or later.

It better. Copyright has essentially fucking ceased to exist in the eyes of AI people. Just because you have a shiny new toy doesn't mean the law suddenly stops applying to you. The internet does its best to route around laws and government but the more technologically up to date bureaucracy becomes, the faster it will catch up.

replies(1): >>safety+6d
◧◩
3. safety+6d[view] [source] [discussion] 2023-12-27 16:30:29
>>jterry+F5
Yeah I mean I'm not even really a fan of how copyright law works, but I don't see how you can just insert an "AI exemption." So OpenAI can infringe because they host an AI tool, but we humans can't? That would be ridiculous. Or is "I used AI when I created this" a defense against infringement? Also seems ridiculous. Why would we legally privilege machine creation of creative works over human creation in the first place? So I don't see what the credible AI-related copyright law reform is going to be yet.

Which means that either OpenAI is allowed to be the only lawbreaker in the country (because rich and lawyers), or nobody is. I say prosecute 'em and tell them to make tools that follow the law.

[go to top]