The concern is about lack of regulation in AI, meaning large private corporations could become more powerful than governments, and that AI agents could (perhaps unwittingly, like Covid) be released. For example, AI is learning to think and act like humans, but does not always take the moral high ground to solve problems it has been set. The leaders in AI are warning us all, but seemingly also do not want to slow down AI development in their country. Also, "rogue" AI agents could hack into energy networks, transport systens, airlines, safety systems, financial institutions, health care, food supply and cause deaths that way. Urgent cross border regulation in AI will help, (but it is happening too slowly). Unregulated Open AI getting into the wrong hands (criminals) could easily be a very dangerous thing, and that is why they are saying that AI could be very dangerous. The future is unpredictable due partly to AI. As with previous challenges to our security, it is up there with other threats (from diseases, war and, of course, last but not least, Climate Change.) It is a force for good, if strictly regulated, used wisely, but it must be regulated asap.