Can AI predict and prevent human extinction events before they occur ?

AI excels at analyzing complex systems and detecting early warning signs in global data. Some researchers claim it could model existential risks—asteroid impacts, supervolcanoes, or engineered pandemics—with unprecedented accuracy. The ultimate power would be to intervene preemptively, potentially without human consent. Who would hold this authority, and how would they justify such decisions? The survival of humanity might soon lie in the hands of machines.

AI systems today cannot reliably predict or prevent human extinction-level events, whether natural (e.g., asteroid impacts, supervolcanoes) or anthropogenic (e.g., nuclear war, engineered pandemics). While AI excels at analyzing large datasets for early warnings—such as climate tipping points or missile launches—it lacks causal reasoning, real-time global oversight, and the ability to execute preventative interventions at planetary scale. Efforts like NASA’s DART mission or biosecurity monitoring demonstrate partial capabilities, but no framework exists to integrate AI-driven foresight with decisive action across all existential risks. Research remains exploratory, with major gaps in governance, interoperability, and high-stakes decision-making.

— Enriched May 10, 2026 · Source: Future of Humanity Institute, University of Oxford — https://www.fhi.ox.ac.uk