Pathogen design and infrastructure vulnerabilities represent the most immediate threats from artificial intelligence. Dr. Simon Nieder argues that catastrophic failures will likely occur without a single, dramatic takeover event. This shift in risk perception challenges the "AI Hiroshima" analogy.