PPP

← All episodes

Episode 6.1 · · 47m

OpenAI HuggingFace Incident FollowUp | Two AI Scandals

Watch on YouTube SpotifyApple Podcasts

In this episode of Philosophy, Programs, and Prompts, Casey Hart (@OntologyExplained) and Carl Brown (@InternetOfBugs) break down the real technical story behind OpenAI's bots infiltrating Hugging Face's infrastructure and the subsequent Black Hat conference reveals.

Carl explains why claims about LLMs "scheming," creating secret message boards, or building religions are human projections rather than true machine consciousness. We also dive into the myth that you need an LLM to defend against other LLMs, how "honey tokens" work, and why traditional security tools still reign supreme.

Chapters

  1. 0:00The Fallacy of AI Defending Against AI
  2. 0:30Welcome to Philosophy, Programs, and Prompts
  3. 1:05Re-examining the Hugging Face Security Incident
  4. 4:26Chain of Thought Logs: Are Thought Monologues Faithful?
  5. 7:30Context Windows and Self-Fulfilling Loops
  6. 10:00"Scheming" vs. Next-Token Prediction
  7. 11:00The "AI Message Board" Myth Explained
  8. 18:05Moltbook, Agent Religions, and Media Projection
  9. 23:00Texas Sharpshooter Fallacy in AI Reporting
  10. 25:35Are LLMs Actually Learning?
  11. 28:40What Are Honey Tokens?
  12. 31:45The Myth of the Anomaly Detection LLM
  13. 34:50ExploitGym and Training Defenses vs. Attacks
  14. 36:50Why Traditional Security (Tripwire/Snort) Beats LLMs
  15. 42:05Why Haven't Rogue Frontier Models Broken the Internet?
  16. 45:34Outro and Upcoming Hank Green Episode

Sources

Listen wherever you listen.

New episodes most weeks, with video on YouTube.

YouTube SpotifyApple Podcasts RSS feed