What’s the BUZZ? — AI in Business Secure Your LLM Against Common Vulnerabilities (Guest: Steve Wilson)
Oct 3, 2023
Steve Wilson, Project Leader at OWASP Foundation, discusses securing Large Language Models (LLMs) against vulnerabilities. Key topics explored include the most important vulnerabilities of LLMs, underestimation of security risks by developers, how these vulnerabilities will be exploited, and mitigation strategies. The podcast emphasizes balancing innovation and security risks, understanding prompt injections, preventing over-assigning agency to LLMs, and establishing trust boundaries with LLM-generated output.
Chapters
Transcript
Episode notes
1 2 3 4 5
Introduction
00:00 • 2min
Conversation and Introduction to 'In Your Own Words' Game
02:00 • 2min
Security Concerns and Risks in AI: Lessons from the Early Days of the Web
04:06 • 9min
New Risks and Vulnerabilities in AI Security
12:47 • 12min
Key Takeaways on Security Considerations of Generative AI
24:50 • 3min