CISO Series Podcast I Taught DeNiro Security Theater, I Can Teach You.
Oct 31, 2023
Davi Ottenheimer, VP of Trust and Digital Ethics at Inrupt, dives into the complexities of security theater in organizations. He explains how security practices often persist due to optics and compliance, even when ineffective. The discussion includes tips for identifying security theater and communicating its risks to non-technical leadership. Davi also highlights the importance of maintaining trust during these transitions and shares insights on ethical AI, the risks of LLMs, and community shifts in the InfoSec landscape.
AI Snips
Chapters
Transcript
Episode notes
Guest's Personal Distrust Of Tesla And Rideshares
- Davi described opting out of rideshare services and distrusting Teslas due to perceived engineering flaws.
- He used strong language to emphasize his personal rejection of certain modern tech.
LLMs Are Confident Mimics, Not Truthful Experts
- Treat current LLMs as confident imitators that often produce plausible but untrue outputs.
- Always plan for human editing or verification because models lack self-awareness and truth-checking.
Always Verify LLM Outputs Externally
- Never rely solely on an LLM to verify facts; always use external validation and domain experts.
- Recognize that refinement can deepen falsehoods unless humans check truth independently.