Wouldn't work: the copyright lawsuit against Anthropic ruled that while copyright infringement (AKA piracy) to get the training data for training an AI is still copyright infringement, actually doing the training once you have the data is not itself a copyright violation.

Simply making the code available, regardless of the licence, allows present and future AI to train on them.

▲

jcarlosweb 7 hours ago | parent | next [-]

But that was in America, where big corporations do whatever they want; in Europe, that shouldn't be allowed. Changing just a few lines in the open-source licenses could save us in the future.

	▲	ben_w 6 hours ago \| parent [-]
		> But that was in America, where big corporations do whatever they want; The ruling was very much not "big corporations do whatever they want", hence where I wrote "while copyright infringement (AKA piracy) to get the training data for training an AI is still copyright infringement" > Changing just a few lines in the open-source licenses could save us in the future. Could? No, not when so much of the coding work gets done in America. Not when the leading AI are American models. Not when so much of the open source code itself is American in origin. Not when the code is hosted on American servers. Now, the code AI makes being bad, that might save our profession, but I'm not counting on it. I've not figured out what to do next, but I don't think I'm staying in software.

▲

re-thc 6 hours ago | parent | prev [-]

> Wouldn't work: the copyright lawsuit against Anthropic ruled that

Then just make it work? Humans invented the laws, the rules, etc.

	▲	ben_w 6 hours ago \| parent [-]
		While laws can indeed be changed, "just" isn't the right word for it. Very few nations empower the electorate to direct binding referendums on topics of the electorate's choice. Switzerland does, I can't think of any others.