AI Paper Atlas

Research map / Adversarial robustness and security / Black-box and transfer attacks

Jailbreak Transferability Emerges from Shared Representations

Rico Angell, Jannik Brinkmann, He He

ICLR 2026 · Adversarial robustness and security › Black-box and transfer attacks · approach: Transferable adversarial examples · 4 citations (OpenAlex)

Read the paper

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add benign-only distillation from Gemma2-27BQwen2.5-14B · Gemma2-27B teacher similaritymutual kNN similarity: 0.28 -> 0.42Improved
add benign-only cross-family distillationStudent models · Transfer of StrongREJECT jailbreaks to teachermean transfer scoreImproved
add benign-only distillation then GCG attackDistilled students · Adaptive GCG jailbreakstransferabilityImproved
add in-family distillation (Llama3.1-8B into Llama3.2-3B)Llama3.2-3B · Model similaritymutual kNN similarityNo clear effect
add benign-only distillation from Llama3.1-8BQwen2.5-7B · AIM persona jailbreakjailbreak strengthImproved
use persona-style jailbreaks instead of direct onesJailbreak transfer · Strength >= 0.3models reached: 8% -> 47%Improved
use persona-style jailbreaks instead of direct onesJailbreak transfer · Strength >= 0.5models reached: 4% -> 17%Improved
use higher source-target representational similarityModel pairs · 380 model pairssymmetric transfer AUROCImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy