Eric Freeman is the CISO at Writer, an AI-native company that has built its own large language model. Before that, he worked across blockchain and emerging technology. Before any of that, he was a line cook pulling 16-hour shifts in a restaurant kitchen six days a week.

That background shows up in everything about how he leads. In this episode, Eric draws a direct line between prepping for dinner service and implementing security controls, between reading a plate and reading a log, between surviving a Friday night rush and surviving a major incident.

We get into how AI is changing both offense and defense in cybersecurity right now, with specific examples of how his team is using LLMs to automate vulnerability validation end-to-end. He explains why context is the only thing that makes AI useful and shares a learning framework where team members use personal analogies to internalize unfamiliar concepts through LLMs.

Eric also doesn't hold back on what's broken. He makes the case that cybersecurity stress is a structural problem, not a personal one, and proposes a mandatory security credit score for businesses. He breaks down prompt injection as social engineering for machines, agents as scripts with more dynamicness, and reduces all of cybersecurity to a single mental model: access control.

We close with his framework for the three camps of cybersecurity buyers, why two of those camps are the reason the industry still sells on fear, and how to build a security culture with engineers by making the secure path the fastest path.

For practitioners, vendors, investors, and anyone trying to understand how the cybersecurity industry actually works underneath the noise.

Connect with Eric Freeman on LinkedIn: https://www.linkedin.com/in/eric-m-freeman/