Siren: A Learning-Based Multi-Turn Attack Framework for Simulating Real-World Human Jailbreak Behaviors

TL;DR

Siren is a learning-based framework that simulates real-world multi-turn jailbreak attacks on large language models, outperforming existing single-turn methods and providing insights for developing stronger defenses.

Contribution

The paper introduces Siren, a novel multi-turn attack framework that uses learning-based strategies to better mimic human jailbreak behaviors in LLMs, surpassing prior static or single-turn approaches.

Findings

Achieves 90% attack success rate against Gemini-1.5-Pro

Attains 70% success against GPT-4o with Mistral-7B attacker

Performs comparably to multi-turn GPT-4o-based attack with fewer turns

Abstract

Large language models (LLMs) are widely used in real-world applications, raising concerns about their safety and trustworthiness. While red-teaming with jailbreak prompts exposes the vulnerabilities of LLMs, current efforts focus primarily on single-turn attacks, overlooking the multi-turn strategies used by real-world adversaries. Existing multi-turn methods rely on static patterns or predefined logical chains, failing to account for the dynamic strategies during attacks. We propose Siren, a learning-based multi-turn attack framework designed to simulate real-world human jailbreak behaviors. Siren consists of three stages: (1) MiniMax-driven training set construction utilizing Turn-Level LLM feedback, (2) post-training attackers with supervised fine-tuning (SFT) and direct preference optimization (DPO), and (3) interactions between the attacking and target LLMs. Experiments demonstrate that Siren achieves an attack success rate (ASR) of 90% with LLaMA-3-8B as the attacker against Gemini-1.5-Pro as the target model, and 70% with Mistral-7B against GPT-4o, significantly outperforming single-turn baselines. Moreover, Siren with a 7B-scale model achieves performance comparable to a multi-turn baseline that leverages GPT-4o as the attacker, while requiring fewer turns and employing decomposition strategies that are better semantically aligned with attack goals. We hope Siren inspires the development of stronger defenses against advanced multi-turn jailbreak attacks under realistic scenarios. Code is available at https://github.com/YiyiyiZhao/siren. Warning: This paper contains potentially harmful text.