The precedent of demonetization, deletion of accounts, and bans, without any evidence was set long ago by Alphabet Inc., who have been doing these things for years, at least since 2016. Untold numbers of creators have been affected, and a jargon has even arisen in the field in order to discuss the triggers and issue without incurring such penalties, as you well understand.

AI is misnamed. LLM's aren't intelligent, or even capable of intelligence at all. They're just mechanisms to weight text samples as more or less probable to be responsive to queries. If a text string they compose contains a formula like 2-2=0, AI has no comprehension of that formula. They don't understand a single word of text. Parrots have some intelligence, and can associate the words they parrot with some meaning. AI has none. Nada. Zip. LLM's have no capacity whatsoever to understand the meaning of the words in text strings.

Consciousness is almost a black hole. It isn't electricity, magnetism, light, or any form of energy we have any theoretical basis in physics for. We can't even directly detect it. The only way we can detect consciousness is from it's secondary effects, when creatures that can move around make conscious decisions. Because trees and barnacles can't move around, we have no way to ascertain if they are conscious or not. Since intelligence underlies conscious decisions, in our very limited understanding of consciousness, intelligence is an aspect of it.

AI isn't conscious, and therefore has no capacity for intelligence, an aspect of consciousness. Because we associate composing text with intelligence and conscious acts, we naturally assume that some composer of text is intelligent. Only our bias lends the appearance of intelligence to AI. Consciousness has been shown in single celled creatures that can learn and make conscious decisions, and single celled creatures do not have brains or nervous systems, that are organs comprised of many cells. So, whatever consciousness is, it doesn't arise in brains or nervous systems. For this reason, neural networks emulating brains aren't liable to produce consciousness. Actual intelligent machines, real AI, potential to become AGI, will only be possible when we understand where consciousness, and therefore intelligence (capacity to understand, an aspect of consciousness), comes from, and build mechanisms that can reproduce whatever facilitates that, whatever it might be.

Since we have no clue what consciousness actually is, we have no idea where it comes from, and haven't even begun to emulate it. AI is improperly named, and is just a play on our bias, that alone lends it the appearance of intelligence. A better name for current AI would be Algorithmic sorting devices, because that's all it is.

Thanks!

Edit: in fact, a better way to understand nervous systems and brains in relation to intelligence is that evolution provided conscious creatures built in LLM's to algorithmically weight environmental factors, because conscious creatures using that information in making conscious decisions preferentially survived to breed.

We are immense colonies of cells, but we feel we are an individual creature. This shows that consciousness, since we can ascertain that single celled creatures also exhibit consciousness, doesn't derive from such colonial megafauna, but can coordinate consciousness across such a colony. This is further reinforced by research that has shown that our gut fauna participate in human decision making. Consciousness is somehow a collective event, even involving multiple species. How and why it entrains myriad living things into an experience we feel is one being is unknown, and further elucidates how little we know about it, what it is and where it comes from.

A good way to understand how AI works is to imagine a device that sorts physical objects by weight. You could supply a device with a stream of circuit boards on a conveyor belt leading to a scale, and since different circuit boards have different weights, the boards could be sorted by their weight. The sorting device has no understanding of what the circuit boards do, and isn't sorting them by their function, merely by their differing mass creating different weight in 1G. You could sort a stream of nuts and bolts the same way, simply changing the weight that triggers the sorting mechanism, and it wouldn't change the ability of the sorting mechanism to understand the function of the items being sorted.

This is why we have AI creating strings of text, images, and sound. It's all based on sorting samples, and is just being applied to different kinds of samples, like the sorting of physical objects by their different weights.