# Learning to Attack and Defend: Adaptive Red Teaming of Language Models via GRPO

arXiv: https://arxiv.org/abs/2606.09701 · HTML: https://mlsanity.com/paper/2606.09701/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- Placed in Reinforcement learning for reasoning, one of 17 directions in LLM reasoning.
- Builds on 32 of its 40 references that are in the map.
- First posted on arXiv 8 Jun 2026.

## Placed in

- LLM reasoning / [Reinforcement learning for reasoning](https://mlsanity.com/lr/reinforcement-learning-for-reasoning/)
- LLM safety and security / [Prompt injection and agent security](https://mlsanity.com/safety/prompt-injection-and-agent-security/)

## Builds on (most cited first, 30 of 32)

- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge](https://mlsanity.com/paper/1803.05457/) (2018)
- [Universal and Transferable Adversarial Attacks on Aligned Language Models](https://mlsanity.com/paper/2307.15043/) (2023)
- [TruthfulQA: Measuring How Models Mimic Human Falsehoods](https://mlsanity.com/paper/2109.07958/) (2021)
- [HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal](https://mlsanity.com/paper/2402.04249/) (2024)
- [Refusal in Language Models Is Mediated by a Single Direction](https://mlsanity.com/paper/2406.11717/) (2024)
- [Red Teaming Language Models with Language Models](https://mlsanity.com/paper/2202.03286/) (2022)
- [Red Teaming Language Models to Reduce Harms: Methods, Scaling Behaviors, and Lessons Learned](https://mlsanity.com/paper/2209.07858/) (2022)
- [XSTest: A Test Suite for Identifying Exaggerated Safety Behaviours in Large Language Models](https://mlsanity.com/paper/2308.01263/) (2023)
- [WildGuard: Open One-Stop Moderation Tools for Safety Risks, Jailbreaks, and Refusals of LLMs](https://mlsanity.com/paper/2406.18495/) (2024)
- [Great, Now Write an Article About That: The Crescendo Multi-Turn LLM Jailbreak Attack](https://mlsanity.com/paper/2404.01833/) (2024)
- [Rainbow Teaming: Open-Ended Generation of Diverse Adversarial Prompts](https://mlsanity.com/paper/2402.16822/) (2024)
- [Generalizing Verifiable Instruction Following](https://mlsanity.com/paper/2507.02833/) (2025)
- [GDPO: Group reward-Decoupled Normalization Policy Optimization for Multi-reward RL Optimization](https://mlsanity.com/paper/2601.05242/) (2026)
- [The Attacker Moves Second: Stronger Adaptive Attacks Bypass Defenses Against Llm Jailbreaks and Prompt Injections](https://mlsanity.com/paper/2510.09023/) (2025)
- [Automated Red Teaming with GOAT: the Generative Offensive Agent Tester](https://mlsanity.com/paper/2410.01606/) (2024)
- [Chasing Moving Targets with Online Self-Play Reinforcement Learning for Safer Language Models](https://mlsanity.com/paper/2506.07468/) (2025)
- [RL Is a Hammer and LLMs Are Nails: A Simple Reinforcement Learning Recipe for Strong Prompt Injection](https://mlsanity.com/paper/2510.04885/) (2025)
- [PyRIT: A Framework for Security Risk Identification and Red Teaming in Generative AI System](https://mlsanity.com/paper/2410.02828/) (2024)
- [Jailbreak-R1: Exploring the Jailbreak Capabilities of LLMs via Reinforcement Learning](https://mlsanity.com/paper/2506.00782/) (2025)
- [Diverse and Effective Red Teaming with Auto-generated Rewards and Multi-step Reinforcement Learning](https://mlsanity.com/paper/2412.18693/) (2024)
- [Learning to Inject: Automated Prompt Injection via Reinforcement Learning](https://mlsanity.com/paper/2602.05746/) (2026)
- [PISmith: Reinforcement Learning-based Red Teaming for Prompt Injection Defenses](https://mlsanity.com/paper/2603.13026/) (2026)
- [TROJail: Trajectory-Level Optimization for Multi-Turn Large Language Model Jailbreaks with Process Rewards](https://mlsanity.com/paper/2512.07761/) (2025)
- [SCOPE-RL: Stable and Quantitative Control of Policy Entropy in RL Post-Training](https://mlsanity.com/paper/2510.08141/) (2025)
- [SEMA: Simple yet Effective Learning for Multi-Turn Jailbreak Attacks](https://mlsanity.com/paper/2602.06854/) (2026)

## Built on by (most cited first, 5 of 5)

- [Trident : How to Break Deep Reinforcement Learning Cyber Defenses (Agentic)](https://mlsanity.com/paper/2608.04317/) (2026)
- [Co-Evolution in Agentic Systems: Toward Self-Directed Evolution Beyond Human Design](https://mlsanity.com/paper/2608.10299/) (2026)
- [ACEA: An Adversarial Co-Evolution Arena for Head-to-Head Red-Team and Blue-Team LLM Testing](https://mlsanity.com/paper/2609.08256/) (2026)
- [CART: Closed-Loop Adaptive Red Teaming for Large Language Models](https://mlsanity.com/paper/2609.27336/) (2026)
- [Red-TTT: Test-Time Training for Automated Jailbreaking Large Language Models](https://mlsanity.com/paper/2610.05282/) (2026)

## Abstract (arXiv)

Language model safety must continually adapt to evolving attacks. Recent works have demonstrated that reinforcement learning can be used to train stronger attacker and defender models in tandem by applying PPO-style self-play and DPO-style online preference optimization. In this work, we explore the efficacy of GRPO in this setting. Co-training can be challenging because it requires jointly optimizing multiple properties of both the attacker and defender. We therefore shape model outputs using multiple LLM judge-based reward channels and compute advantages with GDPO, which prevents any single channel from dominating. Our method uses a curriculum that progresses from attacker-only single-turn and multi-turn training to co-training, where attacker and defender models are updated in alternation. We show that this method produces highly effective and transferable attacks, and that co-trained defenders reach competitive safety while preserving general utility. Through a controlled ablation, we further identify which components of our training pipeline most affect the resulting balance between safety and utility. Finally, we find that GRPO tends to collapse attacker diversity over training and discuss possible ways to address this limitation.