This post on the fediverse goes like this:

that’s why all of the “ethical” LLMs are still trained on, drumroll, huggingface data that was scraped without consent. the people working on this shit don’t consider violating consent and ignoring open source licenses to be unethical.

The whole post here:

This is why I think AI researchers are as morally bankrupt as business owners and their C-suite lackeys. Neither sets of people know anything about ethics, or they purposely ignore their training. And it’s my opinion that those kind of people need to be fed into a woodchipper feet first. No exceptions.

And there’s this too: