AI doomsday scenarios: Researchers warn of extinction risks

Artificial intelligence researchers have long warned that AI could pose existential threats to humanity, including designing unstoppable diseases, triggering nuclear war, or even converting Earth into paper clip factories, though debates over these scenarios' plausibility have intensified amid industry calls to slow development. Skeptics argue such worst-case predictions are sci-fi fantasies that help frontier labs recruit talent, while others like former Anthropic researcher Jacob Coxon insist the risks are real but hard to communicate. RAND Corp. found AI cannot currently cause nuclear war due to safeguards, but warned integration into decision chains could change that. The broader concern centers not on AI acting independently but on humans or nations using the technology to destroy adversaries, with bioweapons development highlighted as a key risk area despite AI's potential for medical breakthroughs.