[whichfawkes]

I'm not so sure about that. It seems to me that they're selling me a service. Just like I might pay for a subscription to Adobe Photoshop or pay per-render fees to a rendering farm.

I could use Photoshop to reproduce a copyrighted work, and in some circumstances (i.e. personal use) that'd be fine. Or I could use Photoshop to reproduce a copyrighted work and try to sell it for profit, which would clearly not be fine. Nobody is saying that Adobe has to recognize whether or not the pixels I'm editing constitute a copyrighted work or not.

[voltaireodactyl]

The difference here is that Adobe is selling a set of tools that can recreate copyrighted work from the ground up. The Mona Lisa being previously incorporated into their tools is not a foundational necessity for their paintbrush to brush digital paint.

The same is not true for AI, which require copyrighted work be contained therein, in order for the tool part to function.

[kuchenbecker]

While I 100% agree, there is another angle to consider this from, in that ChatGPT replaces reading the NYT. ChatGPT competes with it in the delivery of information.

To add to your point though, a sufficiently advanced AI trained on licensed data could reproduce copywrited content from prompt alone. It's the next step that would cause infringement where someone does something withcthe output.