AI Catastrophe

Current trends and expert risk assessments identify catastrophe from advanced artificial intelligence as the single most probable way the world could end. The Doomsday Clock stood at a record 85 seconds to midnight in early 2026, driven by great-power aggression, collapsing arms control, climate extremes, and disruptive technologies including AI. Longer-horizon estimates nevertheless rank unaligned or malicious AI far higher than nuclear war or climate change for causing extinction or the permanent loss of humanity’s future; figures associated with Toby Ord place the century-scale existential risk from unaligned AI near 10 percent, versus roughly 0.1 percent for nuclear war. An October 2026 Lancet Commission likewise found that only malicious AI use and nuclear conflict carry extinction potential this century, with specialists estimating a few percent chance of AI-driven extinction by 2100 and some developers citing substantially higher odds of severe outcomes. Rapid capability growth, the prospect of AI-enabled bioweapons or loss of control, and governance that continues to lag these developments make an AI-driven ending the leading pathway, even while near-term conflicts keep nuclear dangers acute and environmental stresses impose severe but more gradual harm.