Securing AI Systems in Real-World Deployments

Learn about securing AI systems in real-world deployments through this 15-minute DevSecCon conference talk that examines critical security challenges faced when implementing AI in production environments. Explore insights from first-hand experience with the world's largest AI Red Team (Gandalf), covering evolving threat landscapes, prompt injection attacks, data loss prevention, and the security implications of democratized AI capabilities. Discover how increasing AI complexity and autonomy impacts system unpredictability and learn essential strategies for defending against AI-specific threats in production environments.