Perception Point CTO Tal Zamir explains why securing the large language models (LLMs) used to enable generative artificial intelligence (AI) applications is going to be more challenging than most cybersecurity teams realize.