THRD: A Training-Free Multi-Turn Defense Framework for Jailbreak Attacks on Large Language Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ma, Zhiqing, Xu, Zhonghao, Yu, Dong, Kang, Chen, Li, Changliang, Liu, Pengyuan
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914621964681216
author Ma, Zhiqing
Xu, Zhonghao
Yu, Dong
Kang, Chen
Li, Changliang
Liu, Pengyuan
author_facet Ma, Zhiqing
Xu, Zhonghao
Yu, Dong
Kang, Chen
Li, Changliang
Liu, Pengyuan
contents Multi-turn jailbreak attacks pose a growing threat to LLMs by exploiting conversational dynamics such as gradual escalation and cross-turn coordination. Existing defenses either rely on costly retraining -- often degrading model utility -- or apply single-turn analysis independently at each turn, failing to capture how risk accumulates along interaction trajectories. We observe that safety behavior in multi-turn interaction is trajectory-dependent: dialogue history continuously reshapes the model's conditioning context, making it insufficient to evaluate each turn in isolation. Motivated by this insight, we present THRD, the first training-free framework that explicitly models temporal risk accumulation for multi-turn jailbreak defense. THRD integrates four modules: a Turn-level Risk Assessor (TRA) for instantaneous risk estimation, a Historical Context Analyzer (HCA) for cross-turn intent escalation detection, a Response Evaluator (RE) for identifying facilitative outputs, and a Decision Module that combines these signals through a time-evolving scoring mechanism with attenuation-based modulation and trend-aware adjustment. Experiments against state-of-the-art multi-turn attacks -- including tree-search-based and multi-agent collaborative methods -- across two target models show that THRD reduces ASR to 0.2--4.0% while preserving model utility within 1.5% degradation on MMLU and GSM8K. Ablation studies confirm non-redundant module contributions and stable cross-architecture generalization. Analysis of first rejection triggers reveals that over 70% of multi-turn attacks require Turn~2 or later to detect, validating the necessity of explicit temporal aggregation.
format Preprint
id arxiv_https___arxiv_org_abs_2606_01738
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle THRD: A Training-Free Multi-Turn Defense Framework for Jailbreak Attacks on Large Language Models
Ma, Zhiqing
Xu, Zhonghao
Yu, Dong
Kang, Chen
Li, Changliang
Liu, Pengyuan
Computation and Language
Artificial Intelligence
Multi-turn jailbreak attacks pose a growing threat to LLMs by exploiting conversational dynamics such as gradual escalation and cross-turn coordination. Existing defenses either rely on costly retraining -- often degrading model utility -- or apply single-turn analysis independently at each turn, failing to capture how risk accumulates along interaction trajectories. We observe that safety behavior in multi-turn interaction is trajectory-dependent: dialogue history continuously reshapes the model's conditioning context, making it insufficient to evaluate each turn in isolation. Motivated by this insight, we present THRD, the first training-free framework that explicitly models temporal risk accumulation for multi-turn jailbreak defense. THRD integrates four modules: a Turn-level Risk Assessor (TRA) for instantaneous risk estimation, a Historical Context Analyzer (HCA) for cross-turn intent escalation detection, a Response Evaluator (RE) for identifying facilitative outputs, and a Decision Module that combines these signals through a time-evolving scoring mechanism with attenuation-based modulation and trend-aware adjustment. Experiments against state-of-the-art multi-turn attacks -- including tree-search-based and multi-agent collaborative methods -- across two target models show that THRD reduces ASR to 0.2--4.0% while preserving model utility within 1.5% degradation on MMLU and GSM8K. Ablation studies confirm non-redundant module contributions and stable cross-architecture generalization. Analysis of first rejection triggers reveals that over 70% of multi-turn attacks require Turn~2 or later to detect, validating the necessity of explicit temporal aggregation.
title THRD: A Training-Free Multi-Turn Defense Framework for Jailbreak Attacks on Large Language Models
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2606.01738