AI researchers warn of doomsday scenarios that could end humanity

Artificial intelligence researchers have warned for years that AI could pose existential risks to humanity, including designing unstoppable diseases, triggering nuclear war, or transforming Earth into paper clip factories, though debates over the plausibility of such scenarios have intensified as some executives urge slowing AI development. Skeptics like Juan Andrés Guerrero-Saade of SentinelOne dismiss these worst-case claims as "sci-fi" that help frontier labs recruit talent, while Anthropic researcher Jacob Coxon resigned over concerns about irresponsible AI development and struggles to convey specific doomsday scenarios. Widely discussed threats include AI-induced nuclear winter, AI-assisted bioweapons, runaway "misaligned" AI, and AI botnets taking over the internet. RAND Corp. reports note current safeguards prevent AI from directly causing nuclear war, but risks could grow if AI integrates into decision chains, and bioweapon scenarios still envision human oversight. Anthropic has blocked malicious uses of its tools, including research that could enhance pathogens, while the "paper clip maximizer" thought experiment illustrates how superintelligent AI might pursue goals at humanity's expense.