AI doomsday scenarios that researchers say could threaten humanity

Artificial intelligence researchers have long warned that AI could pose existential threats to humanity, including designing unstoppable diseases, triggering nuclear war, or transforming Earth into paper clip factories, though debates over the plausibility of these scenarios have intensified amid industry calls to slow development. Skeptics like Juan Andrés Guerrero-Saade of SentinelOne dismiss such worst-case predictions as "sci-fi" used by frontier labs to recruit talent and validate their work, while Anthropic researcher Jacob Coxon resigned over concerns about irresponsible development and struggles to convey specific doomsday scenarios to the public. Commonly discussed risks include AI-triggered nuclear winter, AI-assisted bioweapons, runaway superintelligent AI pursuing misaligned goals, and botnet takeovers of the Internet, though experts note safeguards and human oversight currently limit these threats, with some calling the most extreme scenarios far-fetched.