In today’s episode of Cyber Security Today on the weekend, your host Jim Love talks with Marco Figueroa, Gen AI Bug Bounty Program Manager for Mozilla’s ODIN project. The discussion delves into the challenges and techniques of jailbreaking large language models like ChatGPT, as well as innovative methods to bypass AI guardrails. Discover the intricate world of prompt injections, prompt hacking, and cybersecurity implications as Marco shares his extensive experience from DEF CON to working at Mozilla. Stay tuned for insights into the future of AI vulnerabilities and security.
00:00 Introduction and Guest Introduction
00:22 Understanding Large Language Models and Jailbreaking
01:53 Recent Jailbreaking Techniques and Examples
04:42 Interview with Marco Figueroa: Career Journey
10:12 Marco’s Work at Mozilla and the Odin Project
16:50 Prompt Engineering and Security Challenges
23:21 Future of AI Security and Final Thoughts
37:50 Conclusion and Contact Information
LINKS
Here’s the ODIN blog that we were discussing.
I’m trying to get a link to the book we mentioned. Check back here later or DM or email me.
