I think its a little dangerous and naive to dismiss the dangers of AI because a this moment in time it's not Skynet. Nobody is saying that ChatGPT is going to destroy humanity, just that there is a huge potential for AI to develop quicker than we can safely control.
Also we shouldn't...
I think a big concern is what happens when AI combines other technology. For example quantum computing. Things like banking encryptions that previously took a 100,000 years to solve can potentially be solved in hours by quantum computers. It makes you wonder what damage a rogue state with and AI...