I think the purpose of AI is to suggest outrageous ideas that no human could get away with suggesting - but first AI has to become established as "all-knowing" and "always correct". Gain a cult-like following, eventually going from "artificial intelligence" to "absolute intelligence", the final arbitrator of all disputes. the final say on every matter.
I think the "AI is dangerous" campaigns are designed to cause people to FEAR AI. To plant the idea into people's minds that AI could become sentient and control our lives... which is by design what it's supposed to do. AI can never become sentient, but we've been programmed to believe that is not only possible, but inevitable. If AI goes "rogue", our overlords have a delightful opportunity for covid-style lockdowns and mandates and restrictions - right out of a dystopian fiction tale. Seems silly now, but not silly when it happens. Just as lockdowns over a cold would have seemed silly in 2010.
AI is not dangerous - it compiles pre-existing information and takes credit itself. It's a sophisticated plagiarizer, stealing the work of people and disguising it as it's own. But since it's not a person, who can make that charge? Any illegal activity done by AI cannot be prosecuted. Are we going to put a computer on trial?