Will the Terminator Imprison Us, or Will Threat Actors Compromise It First? - AI Safety and Security Risk Assessment

Explore AI safety and security risks through actuarial modeling techniques in this 43-minute conference talk from The Diana Initiative 2025. Learn about comprehensive research conducted with top universities over six months to measure the likelihood of AI causing harm versus being compromised by attackers. Discover real-world and science fiction case studies that demonstrate practical approaches to assessing AI vulnerabilities. Gain insights from research that examines whether artificial intelligence systems pose greater risks as autonomous threats or as targets for malicious actors. Understand how actuarial methods can be adapted to evaluate AI security scenarios and quantify potential risks in artificial intelligence deployments.