I mean the allegations that AIs would develop polymorphic malware on their own, or that a LLM trained on the public Internet would make people smart enough to build a nuclear bomb, and so forth along these lines.
Loving the downvote and the hyperbole to ignore the fact that even the current state of "artificial intelligence", as the general public understands it, is still extremely dangerous to society. This doesn't have to escalate into warfare to be a credible threat. Stop living under a rock.
I mean the allegations that AIs would develop polymorphic malware on their own, or that a LLM trained on the public Internet would make people smart enough to build a nuclear bomb, and so forth along these lines.