Securing GenAI End-to-End: From Development to Deployment

Explore a comprehensive approach to securing AI-powered applications throughout the software supply chain in this 19-minute conference talk by Mihai Maruseac from Google. Delve into the challenges of securing software containing AI, addressing executive orders for strengthening supply chains and secure AI development, as well as NIST SP 800-218A guidelines for secure GenAI usage. Learn about securing every step of the AI model training process, from data ingestion to evaluation, including fine-tuning foundational models and model quantization. Discover strategies for securing AI outputs and deployments, and understand the importance of building on stable, secure foundations as the threat landscape evolves in the rapidly advancing AI world.