Episode
"A Theory of Prompt Injection (and why you should study roles)" by Charles Ye, softboiledheart
- Published
- Jun 23, 2026
- Duration seconds
- 1943
- Processing state
not_requested
Actions
POST https://stenobird.com/v1/public/podcasts/lesswrong-curated-popular-5643401/episodes/a-theory-of-prompt-injection-and-why-you-should-study-roles-by-charles-ye-softboiledheart/transcription-requests
Idempotently request low-priority transcript generation for this episode.GET https://stenobird.com/podcast/lesswrong-curated-popular-5643401/a-theory-of-prompt-injection-and-why-you-should-study-roles-by-charles-ye-softboiledheart.md
Read the agent-friendly Markdown representation of this episode resource.
Summary
Summary We've been building a theory of how prompt injections work under the hood.We show it comes down to how LLMs perceive roles (the humble chat template tags).We use this theory to create new attacks, explain some weird mech interp results, and predict when attacks work.We also advocate for a new subfield focused on the science of roles, and sketch some unexplored new research problems.Work supported by CBAI and Cosmos. Another version of this post (with more inline colors) is here, an...