AI Risk
Most Security Experts Say AI Catastrophe Risk Exceeds Their Tolerance
National security practitioners comparing AI catastrophe odds to their own acceptable threshold. IST/FLI AI Risk Barometer survey of U.S. national security practitioners
Most Security Experts Say AI Catastrophe Risk Exceeds Their Tolerance
Why it matters: Among national security experts who quantified both figures, 61% put the probability of an AI-caused catastrophe at or above the maximum risk they would accept to keep pursuing U.S. AI leadership, versus 39% who put estimated risk below their own line.
View source →- Source
- Institute for Security and Technology
- Unit
- % of respondents (n=56)
- Data as of
- July 15, 2026
- Scope
- IST/FLI AI Risk Barometer survey of U.S. national security practitioners