Jailbreaking AGI: Pliny the Liberator & John V on AI Red Teaming, BT6, and the Future of AI Security

In this episode of the Latent Space Podcast, Alessio and Swix interview Pliny the Elder and John V, who delve into the world of AI jailbreaking and security. Pliny explains his focus on universal jailbreaks and the importance of freedom and transparency in AI development, while John V discusses adversarial machine learning and the hacker collective model. They explore the limitations of current safety measures, advocate for open-source data, and share insights into offensive security agents. The conversation covers real-world applications of AI attacks, the challenges of securing AI systems, and the need for a full-stack approach to AI security, emphasizing system-level solutions over model lobotomization. They also highlight the importance of community collaboration and ethical considerations in AI research and development.