Raw LLM Responses

Not exactly sure what kind of protection one might put down with this technology that doesn't result in the larger corporations being able to crush competition. Like, serious talk here. Imagine the protection is "The AI needs to ONLY be trained on Commonwealth images and/or images the artist gave permission for (or that you own the IP for)". The most immediate and easy result is that large corporations (Google, Facebook, Disney, etc) will still have access to brutal swathes of data they already own (Disney in particular owning Marvel and everything they make, for example). Thus, only they can create AI's that can, in turn, generate whatever they want those AI's to generate, foregoing artists in whatever ways they can. Meanwhile, any independent or small business trying to train or use their own AI (there are several opensource models that anyone can set up at their computer for free) to be able to compete in media generation would need to wade through a LOT of expenditures that they can't afford (like, if your everyday artist wants to ensure the images he got are indeed Commonwealth in the countries of origin, that can cost several thousand bucks in fees and time, and that's just the tip of the iceberg). Make something that looks a bit "too close" to what Disney's AI is putting out? They could then sue you into "proving" that the AI you made doesn't have any of their content (The Mouse's army is vast). This is, of course, their goal. It's why they and other media giants backup the "Concept Art Association" which is trying to lobby against AI. So I'm kinda wondering here what sort of "protections" you are talking about outside of the standard "If they create something that's similar enough to something I made, copyright protects me" that's currently in place (After all, as of right now, if anyone uses AI to create a Mickey Mouse, it falls under Disney IP and can get sued if they try to sell the image).