Threat modeling LLM apps
from Practical AI Hosted by Daniel Whitenack and Chris Benson
If you have questions at the intersection of Cybersecurity and AI, you need to know Donato at WithSecure! Donato has been threat modeling AI applications and seriously applying those models in his day-to-day work. He joins us in this episode to discuss his LLM application security canvas, prompt injections, alignment, and more. Sponsors: Assembly AI – Turn voice data into summaries with AssemblyAI’s leading Speech AI models. Built by AI experts, their Speech AI models include accurate speech-to-text for voice data (such as calls, virtual meetings, and podcasts), speaker detection, sentiment analysis, chapter detection, PII redaction, and more. Porkbun – Go to porkbun.com to get .app, .dev, or .foo domain names at Porkbun for only $1 for the first year! Changelog News – A podcast+newsletter combo that’s brief, entertaining & always on-point. Subscribe today. Featuring: Donato Capitella – Website Daniel Whitenack – Website, GitHub, X Show Notes: The Security Canvas for LLM Applications LLM Security Chronicles Playlist on YouTube Should you let ChatGPT control your browser? When your AI assistant has an evil twin Pliny the Liberator - jailbreak guru Upcoming Events: Register for upcoming webinars here!
Embed this episode
Chapters
- 0:00 Welcome to Practical AI
- 0:35 Sponsor: Assembly AI
- 4:10 Donato's journey in gen AI
- 7:10 The most secure LLM
- 8:57 What is a threat model?
- 10:38 Commonplace AI security
- 14:19 Setting up guard rails
- 22:34 Sponsor: Porkbun
- 24:47 Model checking
- 34:17 Closed LLM endpoints
- 37:16 input & output validation
- 41:32 Sponsor: Changelog News
- 43:18 RLHF in alignment
- 46:36 Jailbreakers vs aligners
- 50:24 Exciting things to explore
- 53:24 Thanks for listening!
- 53:51 Outro
No transcript yet.