[ryukoposting]

> Because the fact that it gets scraped and added to a pile of training data to later be monetized really rubs me the wrong way.

I suspect the next AGPL (if not normal GPL) will explicitly prohibit training of closed models against the source material. Ideally it will prohibit training of open weight models too. Either share the entire process, or go piss up a rope.

[danlitt]

If LLM training is found to be a fair use (looks likely), no license will help.

[dotancohen]

My understanding of the GPL is that it currently prohibits the training of LLMs, no need to explicitly mention them. MIT does allow it, however.

[chrka]

Exactly. Just like using LibGen is prohibited.