Amanda CAARSON

Essentially they scrape it from whoever's not blocking their bots. It then becomes part of the model used to train the thing to generate a likely sounding response, or, in this case, images that are likely to meet the criteria specified by the prompt given.

I get what you're saying. But given the way LLMs obtain their data, I'm not sure educating on the beneficial use cases is adequate.

@Pawpower Yeah we can do that. Actually that sounds like a great idea.