“Humanity is predicted to only take serious regulatory action on existential threats from AI after a major catastrophe has already occurred, mirroring the delayed response to global warming.”