Comment by stingraycharles

They are, but nothing bad has happened (apart from OpenAI failing to do basic sandbox engineering), and the “we have achieved recursive self improvement” is widely overblown as something bad or threatening.