Episode

Coercing LLMs to Do and Reveal (Almost) Anything with Jonas Geiping - #678

Podcast
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
Published
Apr 1, 2024
Duration seconds
2907
Processing state
failed
Canonical source
https://twimlai.com/podcast/twimlai/coercing-llms-to-do-and-reveal-almost-anything/
Audio
https://pscrb.fm/rss/p/traffic.megaphone.fm/MLN1896604137.mp3?updated=1711998329
JSON
/v1/public/podcasts/twiml-ai-podcast/episodes/coercing-llms-to-do-and-reveal-almost-anything-with-jonas-geiping-678
Markdown
/podcast/twiml-ai-podcast/coercing-llms-to-do-and-reveal-almost-anything-with-jonas-geiping-678.md

Actions

  • POST https://stenobird.com/v1/public/podcasts/twiml-ai-podcast/episodes/coercing-llms-to-do-and-reveal-almost-anything-with-jonas-geiping-678/transcription-requests
    Idempotently request low-priority transcript generation for this episode.
  • GET https://stenobird.com/podcast/twiml-ai-podcast/coercing-llms-to-do-and-reveal-almost-anything-with-jonas-geiping-678.md
    Read the agent-friendly Markdown representation of this episode resource.

Summary

Today we're joined by Jonas Geiping, a research group leader at the ELLIS Institute, to explore his paper: "Coercing LLMs to Do and Reveal (Almost) Anything". Jonas explains how neural networks can be exploited, highlighting the risk of deploying LLM agents that interact with the real world. We discuss the role of open models in enabling security research, the challenges of optimizing over certain constraints, and the ongoing difficulties in achieving robustness in neural networks. Finally, we delve into the future of AI security, and the need for a better approach to mitigate the risks posed by optimized adversarial attacks. The complete show notes for this episode can be found at twimlai.com/go/678.