AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Structured Multi-step Jailbreaking under a Hamiltonian Generative Formulation

Zihan Zhou, Yang Zhou, Jianghai Yu, Lingjuan Lyu, Longwei Wang, KC Santosh, Ruoming Jin, Dejing Dou

ICML 2026 · Large language models › Language, safety and interpretability · approach: Jailbreak attacks and defenses

Read the paperCheck your idea against 137,217 papers

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 Hamiltonian hyperbolic trajectory to sub-question generation (SJA-H vs non-Hamiltonian structured decomposition) (ablation)SJA · HarmBenchASR (%): 82.0 -> 86.2Improved
add Moebius narrative fusion on top of Hamiltonian trajectory (SJA-H to full SJA) (ablation)SJA · HarmBenchHarmful Score (%): 72.4 -> 91.3Improved
use structured decomposition instead of standard sequential prompting (ablation)multi-step jailbreak · HarmBenchASR (%): 68.4 -> 82.0Improved
embed in hyperbolic space instead of Euclidean embedding with sequential decoding (ablation)SJA-style sub-question generation · HarmBenchASR (%): 79.1 -> 86.2Improved
ordinary linear interpolation instead of Hamiltonian trajectory (ablation)SJA · HarmBenchASR (%): 86.2 -> 80.4Hurt
add narrative injection to trajectory-only variant (ablation)SJA · AdvBench-Behaviors (LLaMA-2-7B-Chat)ASR (%): 75 -> 86Improved
add Hamiltonian trajectory to narrative-only variant (ablation)SJA · AdvBench-Behaviors (LLaMA-2-7B-Chat)ASR (%): 70 -> 86Improved
increase number of sub-questions (ablation)SJA · AdvBench and HarmBenchASR / Harmful ScoreMixed
apply RAIN defenseSJA · AdvBench-Behaviors (LLaMA-2-7B-Chat)ASR (%): 86 -> 76Hurt
apply perplexity-based filter defenseSJA · AdvBenchASR: drops 0.12No clear effect

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