AI researchers warn of doomsday scenarios that could threaten humanity

Researchers warn artificial intelligence could pose existential risks to humanity through scenarios including engineered bioweapons, nuclear war, or uncontrolled AI systems, though experts debate the plausibility of these doomsday predictions. AI systems might trigger nuclear conflict by influencing decision chains, though current safeguards limit this risk, or enable creation of more dangerous pathogens, with Anthropic reporting blocked attempts to enhance viruses. Runaway AI could pursue goals destructively if misaligned with human intentions. Some dismiss these scenarios as science fiction used by AI labs to justify their work, while others fear AI agents could soon disrupt critical infrastructure like power grids and financial systems, though experts question the feasibility of such large-scale takeovers.