Episode 202: Delineating AI Security and Cybersecurity

In episode 202 of Cybersecurity Where You Are, Sean Atkinson and Ed Skoudis sit down with Rob T. Lee, Chief of Research & Chief AI Officer at the SANS Institute. Together, they explore how the implications of multiple 2026 sandbox escapes of artificial intelligence (AI) models are starting to delineate AI security and cybersecurity. Here are some highlights from our episode: 02:00. The historical context of one AI model's 2026 sandbox escape 03:26. The impact of ethics, guardrails, and misconfigurations in an AI containment breach 06:08. Why context matters when talk of AI models "going rogue" surfaces 11:49. How history helps us to delineate AI security and cybersecurity 13:47. A new skill and mindset to match our refined AI risk understanding 15:43. Rules and cybersecurity implementation as a possible way forward 19:04. The double-edged sword of restricting access to open-weight models 23:25. Recommendations for keeping up with what's changing in the AI security space 25:51. Rob's advice: To learn how to defend a thing, try learning how to build it first 28:23. AI governance and seeing through the "slop" to informed conversation 29:54. The use of AI to learn more about and discuss AI security for years to come Resources OpenAI says its AI technology acted on its own in an ‘unprecedented’ hack of another company Pacing model development in an era of cyber-critical capabilities Black Hat USA 2026 | The 'Breaking' News: The OpenAI–Hugging Face Incident Episode 193: AI Security and Responsibility in EO 14409 The AI Daily Brief — Daily AI News & Analysis AI Playbooks for SLTT Cybersecurity Leaders SANS Cybersecurity Summits SANS NewsBites If you have some feedback or an idea for an upcoming episode of Cybersecurity Where You Are, let us know by emailing [email protected].