What is really strange here is that it becomes clear he feeds his conversations into chatgpt to ask what to say in different situations- evidently one that hasn't been jailbroken based on how it reacts to the idea of rape hentai being illegal in the united states and internationally and banned by cloudflare despite the fact that all of those things are legal AND allowed by cloudflare, and then he regurgitates what chatgpt tells him to say, especially when he is highly stressed, based on the verbiage- repeating phrases like 'just go away!' and infinitely replying alongside common keywords like 'boundaries' to AI training makes this really obvious, and it makes me wonder if the feds put all of the docs they had on leto in and said "imitate this person."