# Information Theoretic Adversarial Training of Large Language Models

arXiv: https://arxiv.org/abs/2605.05415 · HTML: https://mlsanity.com/paper/2605.05415/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Jailbreaking, one of 12 directions in LLM safety and security.
- Builds on 15 of its 34 references that are in the map.
- First posted on arXiv 6 May 2026.

## Placed in

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

## Builds on (most cited first, 15 of 15)

- [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)
- [HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal](https://mlsanity.com/paper/2402.04249/) (2024)
- [A General Theoretical Paradigm to Understand Learning from Human Preferences](https://mlsanity.com/paper/2310.12036/) (2023)
- [Latent Adversarial Training Improves Robustness to Persistent Harmful Behaviors in LLMs](https://mlsanity.com/paper/2407.15549/) (2024)
- [Efficient Adversarial Training in LLMs with Continuous Attacks](https://mlsanity.com/paper/2405.15589/) (2024)
- [SelfDefend: LLMs Can Defend Themselves against Jailbreaking in a Practical Manner](https://mlsanity.com/paper/2406.05498/) (2024)
- [Short-length Adversarial Training Helps LLMs Defend Long-length Jailbreak Attacks: Theoretical and Empirical Evidence](https://mlsanity.com/paper/2502.04204/) (2025)
- [Adversarial Training Can Provably Improve Robustness: Theoretical Analysis of Feature Learning Process Under Structured Data](https://mlsanity.com/paper/2410.08503/) (2024)
- [MixAT: Combining Continuous and Discrete Adversarial Training for LLMs](https://mlsanity.com/paper/2505.16947/) (2025)
- [Indirect Gradient Matching for Adversarial Robust Distillation](https://mlsanity.com/paper/2312.03286/) (2023)
- [Improving Accuracy-robustness Trade-off via Pixel Reweighted Adversarial Training](https://mlsanity.com/paper/2406.00685/) (2024)
- [Long-tailed Adversarial Training with Self-Distillation](https://mlsanity.com/paper/2503.06461/) (2025)
- [Uniformly Stable Algorithms for Adversarial Training and Beyond](https://mlsanity.com/paper/2405.01817/) (2024)
- [Explaining the role of Intrinsic Dimensionality in Adversarial Training](https://mlsanity.com/paper/2405.17130/) (2024)

## Abstract (arXiv)

Large language models (LLMs) remain vulnerable to adversarial prompting despite advances in alignment and safety, often exhibiting harmful behaviors under novel attack strategies. While adversarial training can improve robustness, existing approaches are computationally expensive and difficult to scale. Recent continuous adversarial training methods, such as Continuous adversarial training (CAT) and Continuous Adversarial Preference Optimization (CAPO), address this challenge by leveraging gradient-based perturbations in the embedding space, enabling more efficient and expressive attacks. Building on this paradigm, we propose WARDEN, a distributionally robust adversarial training framework for LLMs that dynamically reweights adversarial examples through an f -divergence ambiguity set around the empirical training distribution. Our method optimizes the worst-case adversarial loss within a divergence ball around the empirical data distribution, automatically emphasizing harder adversarial examples. Using the convex dual formulation, the objective reduces to a log-sum-exp form under the KL divergence, with a dynamical parameter controlling the strength of reweighting. This study leads to a new class of information-theoretic objectives that significantly reduce attack success rates while maintaining model utility. Across multiple LLMs and attack settings, WARDEN substantially reduces attack success rates with computational and utility costs comparable to CAT-, CAPO-, and MixAT-based baselines, making it a practical approach for scalable robust alignment.