We have world ending AI, today (nonlineartransform.substack.com)

Recent discussions in the AI community have escalated around the potential dangers posed by advanced AI technologies, with some experts asserting that world-ending threats could materialize today. Contrary to the belief that AI requires physical manifestations like robots to be dangerous, critics argue that the immense computational power available to AI developers currently enables the orchestration of large-scale cyber-attacks. With vast networks of GPUs and AI agents connected to the internet, malicious actors could exploit vulnerabilities in existing cybersecurity infrastructure—often characterized as "laughably bad"—to inflict significant damage on individuals and organizations alike. The implications of such an attack could be catastrophic, leading to widespread societal chaos and destabilization. Scenarios described include the mass corruption of sensitive data, fraudulent legal ramifications for individuals, and disruption of essential services like banking and utilities. This raises urgent questions about our cybersecurity readiness and highlights the need for a comprehensive reevaluation of how AI technologies are safeguarded. As AI capabilities grow, so too does the urgency for the AI/ML community to address these vulnerabilities, emphasizing that the stakes are higher than previously acknowledged.