In the evolving theater of modern warfare, artificial intelligence (AI) has emerged as a transformative force and new frontier of vulnerabilities. This presentation provides a critical analysis of the deceptive behaviors used to turn AI systems into cyber weapons. It is designed to foster technical collaboration throughout the U.S. Department of Defense (DoD) research and engineering (R&E) community.
Key topics for discussion include:
- An analysis of AI foundational vulnerabilities, including poisoned training data, adversarial inputs, and model stealing.
- The devastating potential of AI exploits critical infrastructure, financial markets, and the integrity of information through propaganda.
- Weaponizing AI: methodologies for using AI to hack and subvert existing AI systems.
- A proactive defense strategy rooted in a Secure AI Development Life Cycle, from data collection to deployment.
- Dedicated AI red teaming to expose vulnerabilities before they can be exploited by adversaries.
This presentation will be purely scientific and technical. The content, drawn from Professor Kurtz’s research, directly applies to DoD R&E topics.