Luckily, the legal space here seems to be very congruent with the idea that you do not own works you create with "AI"-enhancements, and thus there's less of an incentive to publish such knowingly.
That is not what current US law states anywhere. There are huge grey areas, but what people seem to be misconstruing is a 2021 ruling that AI
without any significant human input cannot be copyrighted under U.S. law. That leaves a
vast amount of grey area, and the USCO has further specified, this year, that work created with AI
absolutely may be copyrighted if there is "sufficient" human creativity.
In other words, if you just type in the prompt: "do an image of a magic-infused kaiju-T-rex with multiple limbs," the result almost certainly
is not copyrightable. In the US. If you actually create such a painting and then ask an AI to enhance, it almost
is certainly copyrightable. In the US (the US: not the centre of the universe. Other countries have their own laws). Again, creatives have been using various types of AI-assisted enhancement on all kinds of art for decades now, without issue.
So you are badly misunderstanding the current state of US law on AI enhancement. These laws are still being created and there is plenty of litigation ongoing, and we have probably barely scratched the tip of the iceberg. But there absolutely is not some giant legal prohibition against using any form of AI in your creative work in the US, or mandating that you lose copyright protection the instant you use AI. WotC likely didn't drop this artist because of legal worries, they dropped him because of PR.
Note: not a lawyer, but I checked with my bestie who is a lawyer with decades of experience, though not in the US (New York State) since the early 2000s; he now practices in Canada. I wish someone like Snarff would weigh in. My buddy also pointed out that his law firm has been using AI-generated text on many legal documents for some years now, saving quite a bit of money by doing so.