# TROJail: Trajectory-Level Optimization for Multi-Turn Large Language Model Jailbreaks with Process Rewards

arXiv: https://arxiv.org/abs/2512.07761 · HTML: https://mlsanity.com/paper/2512.07761/

## Key facts

- Cited by 14 papers in the map (data through 14 Oct 2026).
- Placed in Jailbreaking, one of 12 directions in LLM safety and security.
- Builds on 36 of its 55 references that are in the map.
- First posted on arXiv 8 Dec 2025.

## Placed in

- LLM safety and security / [Jailbreaking](https://mlsanity.com/safety/jailbreaking/)

## Builds on (most cited first, 30 of 36)

- [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://mlsanity.com/paper/1907.11692/) (2019)
- [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://mlsanity.com/paper/2307.09288/) (2023)
- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Universal and Transferable Adversarial Attacks on Aligned Language Models](https://mlsanity.com/paper/2307.15043/) (2023)
- [Gemma 2: Improving Open Language Models at a Practical Size](https://mlsanity.com/paper/2408.00118/) (2024)
- [HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal](https://mlsanity.com/paper/2402.04249/) (2024)
- [Jailbreaking Black Box Large Language Models in Twenty Queries](https://mlsanity.com/paper/2310.08419/) (2023)
- [Red Teaming Language Models with Language Models](https://mlsanity.com/paper/2202.03286/) (2022)
- [MPNet: Masked and Permuted Pre-training for Language Understanding](https://mlsanity.com/paper/2004.09297/) (2020)
- [How Johnny Can Persuade LLMs to Jailbreak Them: Rethinking Persuasion to Challenge AI Safety by Humanizing LLMs](https://mlsanity.com/paper/2401.06373/) (2024)
- [GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher](https://mlsanity.com/paper/2308.06463/) (2023)
- [Great, Now Write an Article About That: The Crescendo Multi-Turn LLM Jailbreak Attack](https://mlsanity.com/paper/2404.01833/) (2024)
- [RAGEN: Understanding Self-Evolution in LLM Agents via Multi-Turn Reinforcement Learning](https://mlsanity.com/paper/2504.20073/) (2025)
- [A Wolf in Sheep's Clothing: Generalized Nested Jailbreak Prompts can Fool Large Language Models Easily](https://mlsanity.com/paper/2311.08268/) (2023)
- [MiniLMv2: Multi-Head Self-Attention Relation Distillation for Compressing Pretrained Transformers](https://mlsanity.com/paper/2012.15828/) (2020)
- [ArCHer: Training Language Model Agents via Hierarchical Multi-Turn RL](https://mlsanity.com/paper/2402.19446/) (2024)
- [ArtPrompt: ASCII Art-based Jailbreak Attacks against Aligned LLMs](https://mlsanity.com/paper/2402.11753/) (2024)
- [Rainbow Teaming: Open-Ended Generation of Diverse Adversarial Prompts](https://mlsanity.com/paper/2402.16822/) (2024)
- [AutoDAN-Turbo: A Lifelong Agent for Strategy Self-Exploration to Jailbreak LLMs](https://mlsanity.com/paper/2410.05295/) (2024)
- [Free Process Rewards without Process Labels](https://mlsanity.com/paper/2412.01981/) (2024)
- [LLM Defenses Are Not Robust to Multi-Turn Human Jailbreaks Yet](https://mlsanity.com/paper/2408.15221/) (2024)
- [X-Teaming: Multi-Turn Jailbreaks and Defenses with Adaptive Multi-Agents](https://mlsanity.com/paper/2504.13203/) (2025)
- [Dense Reward for Free in Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2402.00782/) (2024)
- [STAIR: Improving Safety Alignment with Introspective Reasoning](https://mlsanity.com/paper/2502.02384/) (2025)
- [Direct Multi-Turn Preference Optimization for Language Agents](https://mlsanity.com/paper/2406.14868/) (2024)
- [Safety in Large Reasoning Models: A Survey](https://mlsanity.com/paper/2504.17704/) (2025)
- [Learning diverse attacks on large language models for robust red-teaming and safety tuning](https://mlsanity.com/paper/2405.18540/) (2024)
- [SPEED: Scalable, Precise, and Efficient Concept Erasure for Diffusion Models](https://mlsanity.com/paper/2503.07392/) (2025)
- [MRJ-Agent: An Effective Jailbreak Agent for Multi-Round Dialogue](https://mlsanity.com/paper/2411.03814/) (2024)
- [MTSA: Multi-turn Safety Alignment for LLMs through Multi-round Red-teaming](https://mlsanity.com/paper/2505.17147/) (2025)

## Built on by (most cited first, 14 of 14)

- [Rethinking Jailbreak Detection of Large Vision Language Models with Representational Contrastive Scoring](https://mlsanity.com/paper/2512.12069/) (2025)
- [TRACE: Task-Aware Adaptive Self-Evolving Agentic Jailbreaking](https://mlsanity.com/paper/2605.30883/) (2026)
- [Learning to Attack and Defend: Adaptive Red Teaming of Language Models via GRPO](https://mlsanity.com/paper/2606.09701/) (2026)
- [Not All Turns Matter: Credit Assignment for Multi-Turn Jailbreaking](https://mlsanity.com/paper/2605.08778/) (2026)
- [Internalizing Safety Understanding in Large Reasoning Models via Verification](https://mlsanity.com/paper/2605.08930/) (2026)
- [Plant, Persist, Trigger: Sleeper Attack on Large Language Model Agents](https://mlsanity.com/paper/2605.28201/) (2026)
- [SoK: Intent-Oriented Systematization of Multi-Turn LLM Jailbreaks](https://mlsanity.com/paper/2608.01117/) (2026)
- [PsychJail: Exploring Psychological Jailbreaks via Multi-Turn Persuasion of LLM Policies](https://mlsanity.com/paper/2608.23028/) (2026)
- [Model-Agnostic Lifelong LLM Safety via Externalized Attack-Defense Co-Evolution](https://mlsanity.com/paper/2605.13411/) (2026)
- [MJ: Multi-turn LLM Jailbreaking via Decomposed Credit Assignment](https://mlsanity.com/paper/2607.11070/) (2026)
- [In-Context Learning as Implicit Policy Gradient](https://mlsanity.com/paper/2607.23153/) (2026)
- [TRACE: Trajectory Aware Reasoning for Multi-Turn Adversarial Conversation Evaluation](https://mlsanity.com/paper/2608.15594/) (2026)
- [HASTE: Evolving Agent Harnesses Against Emerging Attacks Using Sparse Evidence](https://mlsanity.com/paper/2610.02920/) (2026)
- [ATTCK-R: Knowledge-Aware Long-Horizon Jailbreaking for Protected LLMs](https://mlsanity.com/paper/2610.16305/) (2026)

## Abstract (arXiv)

Large language models have seen widespread adoption, yet they remain vulnerable to multi-turn jailbreak attacks, threatening their safe deployment. This has led to the task of training automated multi-turn attackers to probe model safety vulnerabilities. However, existing approaches typically rely on turn-level optimization, which is insufficient for learning long-term attack strategies. To bridge this gap, we formulate this task as a multi-turn reinforcement learning problem, directly optimizing the harmfulness of the final-turn response as the outcome reward. To address the sparse supervision of the outcome reward, we introduce TROJail, which employs two process rewards to evaluate the utility of intermediate prompts and integrate them into advantage estimation. These rewards (1) penalize overly harmful prompts that trigger the model's refusal mechanism, and (2) encourage steering the semantic relevance of responses toward the targeted harmful content. Experimental results show improved attack success rates across multiple models and benchmarks, highlighting the effectiveness of our approach. The code is available at https://github.com/xxiqiao/TROJail. Warning: This paper contains examples of harmful content.