Inception in DiffusionGemma - Jailbreaking a Diffusion Language Model by Pinning Tokens Anywhere on the Canvas

·LessWrong··

Authors: Theresa G., Simon S., Siva Kumar Lakkoju.Epistemic status/effort: exploratory red-teaming as part of a two-day hackathon during ARENA 8.0. Our attacks can be reproduced based on our GitHub Repo (here). The interpretations are mostly intuitions, as the evaluation was small. Hence, treat the framing as "this attack surface is worth taking seriously" in the context of diffusion models moving into production for things like inline editing.TL;DRAuto-regressive LLMs generate from left to righ...

Read full article →

Related Articles

Alphabet's cash burn raises alarm for Big Tech as AI spending climbs
1vuio0pswjnm7 · Hacker News · 10h ago
DARPA, U.S. Air Force fly AI-controlled F-16
r2sk5t · Hacker News · 9h ago
Show HN: Echo – Fable-level results at 1/3 the cost using open-weight models
adam_rida · Hacker News · 4h ago
Everyone should know SIMD
WadeGrimridge · Hacker News · 1d ago
LG to ban residential proxies from smart TV apps
DemiGuru · Hacker News · 1d ago