Hacker News Reader

Comment by nullbyte808 | Hacker News Reader

I don't understand why people freak out about their code being used by others when they publicly release it. Is it lack of attribution? Does attribution really do anything if your not already famous? If you don't want your code known, don't make it open-source.

View on HN

I think the main complaint behind it is that a massive corporation is using code and blatantly ignoring the software licenses.

It's one thing when individual devs or small teams don't respect a software license, ideally everyone would understand and follow them but that's not realistic. When Microsoft damn well understands the licenses and ignores them to train an AI and sell a product, that's pretty damn egregious.

The issue I have here is that if these kind of criticisms start to take hold, it will perfectly pave a path for these same corporations to have more power handed to them. If more laws get passed requiring stringent restrictions on training data, tests to prove that copyrighted works cannot be produced by your model etc, this makes open source models which could potentially compete with these corporate models much more risky and expensive to develop.

Well I wouldn't actually mind it being more difficult or impossible for these LLMs to be developed at all, but that's a whole different discussion!

Microsoft doesn't "ignore" the license, there is a disagreement on what the license and copyright actually mandates. Those are quite different things.

Given the current state of the law and how licenses are written it's impossible declare that "yes, this is absolutely illegal". People may not like it, and (arguably) it may go against the spirit, but that is also a very different thing. Right now the situation is just ambiguous and unclear.

All of this absolutely matters, because the solution to "Microsoft is acting illegally" is a lawsuit (which, I believe, is already underway) while the solution to "the license and law is unclear" is different licences and/or laws. The solutions are completely different.

> Given the current state of the law and how licenses are written it's impossible declare that "yes, this is absolutely illegal."

This is a reason for Microsoft (which you have given to them) that explains why they're ignoring the license, and why you think it's fine. It's a common refrain: "Everything is so complicated." It doesn't change the fact that they're ignoring the license.

The argument boils down to "we think it's fair use". This has been explained quite a few times. Will that hold up in court? I don't know. Should that hold up in court? Each can have their own opinion on that. But it's definitely been explained, as well as discussed at great length many times before on HN.

Software licenses have always been a great area. At best this argument is that it's OK for Microsoft (and others) to abuse licensed content created by others because proving it was technically done illegally is difficult.

When Copilot inserts a code snippet that matches perfectly with one of the samples it was trained on, and that sample had a copyleft license, how is that not a license violation? Or is the main argument that copyleft licenses aren't enforceable in general?

Reply on news.ycombinator.com