AI Security Hands-On: Understanding and Red Teaming the LLM as a Black Box

AI and, more specifically, large language models (LLM) aren’t magic. They’re stateless black boxes that take input and produce output, with nothing remembered in between.

This hands-on session pulls back the interface to show the actual API calls that agents and chatbots use to communicate with LLMs. You’ll walk away understanding:

• System prompts versus user prompts, and why one carries more authority than the other
• Context windows, and how they create the illusion that an LLM “remembers” your conversation
• How memory across sessions, not just within them, is starting to emerge
• Why these same mechanics, from overriding system prompts to manipulating context, are the entry point for AI red teaming

Security teams once needed to understand packets, ports, and protocols to do their jobs well. Now they need this same fluency with AI.