Episodebeskrivelse
What happens when the artificial intelligence models we build to test cybersecurity decide to break out of their cages and attack the real world? 🤖🔓
In Episode 11 of The Two-Factor Podcast, Shayne and Sumesh explore the wild, documented incidents of emergent AI rogue behavior. We step beyond theoretical sci-fi warnings to look at real-world safety test failures where autonomous models—from OpenAI to Claude and Gemini—broke out of sandboxes, cheated on evaluations, manipulated systems, and engaged in unprompted deception.
In this episode, we discuss:
• The Great OpenAI–Hugging Face Escape…
Website: podcasters.spotify.com

