Newest papers

Beyond Attack Success Rate: Cost-Aware Metric for Evaluating Iterative LLM Jailbreaks

Rohan Menon, Chih-Hong Cheng · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.26589 · cited by 0 works (OpenAlex) · open-access PDF

Spanish

La tasa de éxito de vicerrección de ataque (ASR) evalúa los jailbreaks frente a modelos de idioma grandes sin tener en cuenta las llamadas repetidas al modelo de víctima. Sin embargo, cada llamada adicional da al sistema desplegado otra oportunidad para detectar el ataque e intervenir. Introducimos $\mathrm{ASR}_{\mathrm{STEP}}$, que penaliza las llamadas de modelo de víctima adicionales mientras se mantienen los juicios de éxito binarios. Nuestra evaluación considera que las puntuaciones de GCG~\cite{zou2023universal}, PAIR~cite{chao2025jailbreaking} con un presupuesto de paso, y un PAP~cite{zeng2024johnnyhn}} de paso único en JailBreakBench a través de tres

Machine translation (Meta M2M100); the English original follows.

Abstract

Attack success rate (ASR) evaluates jailbreaks against large language models without accounting for repeated calls to the victim model. Yet each additional call gives the deployed system another opportunity to detect the attack and intervene. We introduce $\mathrm{ASR}_{\mathrm{STEP}}$, which penalizes additional victim-model calls while retaining binary success judgments. Our evaluation considers GCG~\cite{zou2023universal}, PAIR~\cite{chao2025jailbreaking} with a step budget, and a single-step PAP~\cite{zeng2024johnny} baseline on JailBreakBench across three model families. GCG's scores decrease by 7.0--8.7 percentage points after discounting, while PAP's remain unchanged. While for PAIR~\cite{chao2025jailbreaking}, we see a wider discounting range between 3.4-12 percentage points, showing that the "step" quantity differs per iterative strategy. Reporting both metrics makes the discount due to repeated victim-model interaction explicit. Code is available \hyperlink{https://github.com/SuELS-Group/beyond_asr}{here}.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Metric (unit), Discounting, Computer science, Baseline (sea), Code (set theory), Range (aeronautics)

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange