In this report, RAND researchers examine the threat of human extinction posed by artificial intelligence (AI). Using a scenario-based analysis to evaluate three technologies — nuclear weapons, pathogens, and geoengineering — they found that it would be immensely challenging for AI to create an extinction threat, although they could not rule out the possibility. They identify risk indicators and make recommendations to mitigate future risk.

On the Extinction Risk from Artificial Intelligence

Michael J. D. Vermeer, Emily Lathrop, Alvin Moon

ResearchPublished May 6, 2025

The capabilities of artificial intelligence (AI) have accelerated to the point at which some experts are advocating that it be taken seriously as a credible threat to human existence. In this report, RAND researchers examined and seriously considered the threat of human extinction posed by artificial intelligence (AI). They used an exploratory scenario-based analysis to evaluate AI's ability to create an extinction threat with three technologies: nuclear weapons, pathogens, and geoengineering. For each scenario, they assessed that it would be immensely challenging to create an extinction threat, although they could not rule out the possibility. They also concluded that an extinction threat would take a significant amount of time to take effect in each scenario, likely giving humans time to respond to and mitigate threats. From their analysis, they identified specific indicators of risk arising from each of the scenarios and follow with a series of recommendations to mitigate extinction risk from AI, including the need to study AI risks more broadly, to build human resilience, and to analyze other technologies that might increase risk. They conclude the study with a research agenda that will help reduce uncertainties and aid efforts to understand and mitigate extinction risk.

Key Takeaways

  • In the three scenarios examined in this study — nuclear weapons, pathogens, and geoengineering — human extinction would not be a plausible outcome unless an actor was intentionally seeking that outcome. Even then, an actor would need to overcome significant constraints to achieve that goal.
  • Analysis under uncertainty requires specific analytic approaches.
  • Extinction threats posed by AI are immensely challenging but cannot be ruled out.
  • Extinction threats occur over long timescales, allowing time to respond.
  • AI would require four capabilities to create extinction threats: (1) integration with key cyber-physical systems, (2) the ability to survive without human maintainers, (3) the objective to cause human extinction, and (4) the ability to persuade or deceive humans to avoid detection.

Recommendations

  • Continue to perform AI risk research but widen the focus to include other risks in addition to extinction risk. These include global catastrophic risks, potential human disempowerment, and AI safety and equity concerns.
  • Improve human resilience in the face of potential global catastrophes with such policy measures as nuclear nonproliferation policy, pandemic preparedness, and agreements like the Montreal Protocol and the Kigali Amendment. Continue efforts for both existing and emerging threats.
  • Focus research and analysis on technologies that would mediate extinction risk. Focus on specific technologies that a highly capable AI actor might use to cause harm.
  • Evaluate known and emerging threats, including those involving deep uncertainty or recognized ignorance.
  • Monitor existing risk indicators and identify other risk indicators.
  • Perform research and craft policy that will shorten the time to decision and time to action. Establish clear circumstances that would trigger a response and clear actions when triggers are observed while recognizing that, although some consequences might occur quickly, true extinction threats develop over longer timescales.

Topics

Document Details

Citation

Chicago Manual of Style

Vermeer, Michael J. D., Emily Lathrop, and Alvin Moon, On the Extinction Risk from Artificial Intelligence. Santa Monica, CA: RAND Corporation, 2025. https://www.rand.org/pubs/research_reports/RRA3034-1.html.
BibTeX RIS

Research conducted by

This work was independently initiated and conducted within the Technology and Security Policy Center of RAND Global and Emerging Risks using income from operations and gifts from philanthropic supporters. A complete list of donors and funders is available at www.rand.org/TASP.

This publication is part of the RAND research report series. Research reports present research findings and objective analysis that address the challenges facing the public and private sectors. All RAND research reports undergo rigorous peer review to ensure high standards for research quality and objectivity.

This document and trademark(s) contained herein are protected by law. This representation of RAND intellectual property is provided for noncommercial use only. Unauthorized posting of this publication online is prohibited; linking directly to this product page is encouraged. Permission is required from RAND to reproduce, or reuse in another form, any of its research documents for commercial purposes. For information on reprint and reuse permissions, please visit www.rand.org/pubs/permissions.

RAND is a nonprofit institution that helps improve policy and decisionmaking through research and analysis. RAND's publications do not necessarily reflect the opinions of its research clients and sponsors.