This talk provides an in-depth look at the key security risks in LLM-based systems and applications. I will explore threats such as prompt injection, data poisoning, model leakage, and other vulnerabilities, explaining their impact, demonstrating real-world examples, and discussing effective mitigation strategies.

Attendees will gain practical insights into securing LLM implementations across different use cases, from customer-facing applications like chatbots and AI assistants to internal tools and deeply integrated AI-driven processes. This session is designed for technical and leadership roles alike, equipping them with the knowledge to proactively address security risks before they become critical threats.