There’s a fallacy that gets used a whole lot to justify things like this (not just with LLMs), and I see it in many of the comments here: If it’s OK (or at least negligible on a small scale), then it must be OK on a large scale.

It usually goes something like: If I can make money by learning something from a web page, why does a computer making money by learning everything from everyone upset people so? It’s the same thing!

It’s like if I go to Golden Gate Park and pick one flower, I shouldn’t do that, but no one cares. But if I build a machine to automatically cut every flower in the park because I want to sell them, that’s different.

“You say I can pick one flower, but you get upset when I take a bunch. That’s inconsistent. Check and mate.”

But quantitative changes in an activity produce qualitative changes. Everyone knows this, but sometimes they seem to find it inconvenient to admit it. Not that effects of the qualitative change are always bad, but they are often different, and worth considering rather than dismissing.