A major nonprofit policy research organization that, through its Center on AI, Security, and Technology (CAST) and Global and Emerging Risks division, conducts influential research on AI safety, frontier model security, AI governance, and existential risk policy.
A major nonprofit policy research organization that, through its Center on AI, Security, and Technology (CAST) and Global and Emerging Risks division, conducts influential research on AI safety, frontier model security, AI governance, and existential risk policy.
People
Updated 05/18/26President & CEO
Senior Vice President and Chief Research Officer
Vice President and Chief Human Resources Officer
Vice President and Chief Information Officer
Social and Economic Policy Advisory Board Member
Member, President’s Impact Forum
Adjunct Senior Researcher and Advisor
Research Fellow; Adjunct Staff
Funding Details
Updated 05/18/26- Annual Budget
- $461,500,000
- Current Runway
- -
- Funding Goal
- -
- Funding Raised to Date
- -
Org Details
Updated 05/18/26The RAND Corporation is a nonprofit, nonpartisan research organization founded on May 14, 1948, when it was incorporated as an independent entity after originating as Project RAND in 1946 under the U.S. Army Air Forces. Headquartered in Santa Monica, California, with offices in Arlington (Virginia), Pittsburgh, Boston, and international locations in Cambridge (UK), Brussels, and Canberra, RAND employs approximately 1,850 people and operates with annual revenues of roughly $460 million. Its mission is to help improve policy and decisionmaking through research and analysis. RAND's AI safety and emerging technology work is housed primarily in the Global and Emerging Risks division and its flagship Center on AI, Security, and Technology (RAND CAST), founded and directed by Sella Nevo, who previously led a team of 70 researchers at Google working on machine learning for humanitarian and climate challenges. RAND CAST was built on the work of two earlier centers: the Technology and Security Policy Center (TASP) and the Meselson Center for Biodefense. RAND CAST focuses on several critical areas. In AI security, the center has produced landmark research on securing frontier AI model weights, providing a detailed playbook for frontier AI labs to benchmark and strengthen their security measures. In AI governance, RAND has identified four practicable governance approaches spanning federal regulation to voluntary industry partnerships. The center also studies AI loss-of-control scenarios, analyzing warning signs such as deception, self-preservation, and autonomous replication capabilities in advanced AI models. In biosecurity, RAND evaluates how frontier large language models may enable biological or chemical weapons development. In October 2024, The Audacious Project committed approximately $38 million to RAND and METR for Project Canary, a research collaboration aimed at developing tools that all frontier AI developers can use to evaluate their systems for risks before release. By 2027, Canary aims to enable rigorous safety evaluations for any new AI system prior to deployment. The project is led by RAND researcher Ella Guest and METR founder Beth Barnes, who have already evaluated cutting-edge models for OpenAI and Anthropic. RAND's CEO Jason Matheny brings deep expertise in existential risk. Before joining RAND in July 2022, he served as founding director of Georgetown's Center for Security and Emerging Technology, director of the Intelligence Advanced Research Projects Activity (IARPA), and director of research at the Future of Humanity Institute at Oxford University. He has co-led the National AI R&D Strategic Plan and served on White House committees on AI, biosecurity, and quantum information science. RAND has received significant philanthropic support for its AI safety work. Open Philanthropy has made multiple grants totaling over $10 million for emerging technology initiatives, fellowships, and research. The Survival and Flourishing Fund recommended a $1 million grant (plus matching) to RAND's Technology and Security Policy Center in 2025. RAND has published influential reports on AI extinction risk, AI and critical infrastructure, frontier model biological capability benchmarking, and emergency preparedness for AI loss-of-control incidents.
Theory of Change
Updated 05/18/26RAND's theory of change for AI safety operates through providing rigorous, nonpartisan policy analysis to decision-makers in government and industry at a critical juncture in AI development. By producing technically grounded research on frontier AI risks -- including model weight security, loss-of-control scenarios, biological misuse potential, and governance frameworks -- RAND aims to give policymakers actionable options for managing AI risks before catastrophic capabilities emerge. Through Project Canary, RAND and METR are building standardized pre-release safety evaluation tools that could become industry-wide infrastructure for risk assessment. RAND's unique position as a trusted, long-established policy institution gives its AI safety research credibility with government officials, military leaders, and industry executives who might not engage with newer or more advocacy-oriented organizations. By training the next generation of technology policy analysts through its CAST fellowship program and Pardee RAND Graduate School, RAND also aims to build lasting institutional capacity for AI governance.
Grants Received
Updated 05/18/26Projects
Updated 05/18/26
RAND Global and Emerging Risks is a research division of RAND Corporation delivering rigorous, objective public policy research on catastrophic and existential risks to civilization, including AI, synthetic biology, climate, and nuclear threats.
RAND’s Technology and Security Policy Center (TASP) analyzes how high-consequence, dual-use technologies such as artificial intelligence and biotechnology affect global competition and security, and develops policy and technological solutions to strengthen the security and competitive advantage of the United States, its allies, and partners.
Discussion
Key risk: RAND’s evaluations and governance frameworks (e.g., Project Canary) could be co-opted into permissive box-checking standards that safety-wash risky deployments, with marginal philanthropic dollars diluted inside a large, multi-mission bureaucracy.
Case for funding: RAND is uniquely positioned to translate technically grounded AI risk work—especially standardized pre-release evaluations via Project Canary and model-weight security playbooks—into policies and practices actually adopted by U.S. agencies and frontier labs, materially boosting state capacity to manage AI x-risk.