You say this because you think you understand copyright law. If you actually knew anything about copyright law, you'd never say this. Nobody who understands copyright law thinks it's been done right, unless they're getting rich off of it.
Scraping data has been allowed for decades. It's the foundation of image search engines. We allowed large-scale image scraping and categorization this whole time because we liked the results. Now that there are results we don't like, we have a lot of back-pedaling to do if we want something different. New laws would need to be written to reign this in, and those laws might end up destroying the efficacy of image search engines in the process.
As understandably upset that artists get that AI "steals their style", existing copyright law allows me, without an AI, to steal anyone's style that I want to, because artistic style cannot be copyrighted. If you want to protect artistic styles from being stolen by an AI, you need new laws to protect styles because they don't currently exist at all. Those laws might end up having a chilling effect on things like parody and satire if aesthetics can be owned and protected.
And this is just arguing against the ways the system isn't, as you claim, already prepared to handle the concerns surrounding AI. There are countless other shortcomings. The entire system is broken, partly because it was conceived pre-Internet and hasn't aged well into the modern age, but mostly because it protects giant corporations above all, so remember that when you're begging it to protect small artists from big tech companies.