DACE: Diversity-Driven Adversarial Co-Evolution for Robust LLM Safety Alignment
Ensuring robust safety alignment of large language models (LLMs) is increasingly difficult as adversarial attacks evolve and outpace alignment pipelines built on pre-collected data. Existing co-evolutionary frameworks let the defender chase a moving attacker, yet their training dynamics expose two compounding pathologies: \emph{attack strategy collapse}, where the attacker overfits to a narrow set of high-reward rewrites, and \emph{defense adversarial forgetting}, where the defender loses competence against earlier attacks as the attack distribution drifts. Existing remedies are partial: attacker-side diversity rewards score textual novelty rather than strategy novelty, and defender-side replay relies on discrete judge scores that ignore estimation uncertainty, response stochasticity, and defender non-stationarity. We introduce \textbf{\method}, a diversity-driven adversarial co-evolution framework that targets both pathologies jointly. On the attacker side, \method couples an explicit two-dimensional strategy space (risk category $\times$ attack style) with a \emph{normalized marginal coverage gain} reward, providing a bounded, non-vanishing exploration signal at the strategy layer. On the defender side, \method maintains a unified \emph{Bayesian adversarial replay pool} whose Beta--Bernoulli threat posteriors are refreshed by time decay and sampled via Thompson sampling, anchoring defender training to an evolving threat landscape. Across standard safety, automated-attacker, and general-capability benchmarks, \method improves robustness to out-of-distribution attacks while preserving general capabilities, and yields broader attacker strategy coverage.
Authors
- Zhida He
- Chaochao Lu
- Peng Fei Yu (ORCID: https://orcid.org/0000-0001-9986-0438)
- Han Qi (ORCID: https://orcid.org/0009-0005-9348-0342)
- Qiaosheng Zhang
- Ying Wen
- Ziyuan Zhou
- Shao Zhang
- Xiaoyu Wen
Publication Details
- Journal
- arXiv (Cornell University)
- Published
- 2026-11-17
- DOI
- https://doi.org/10.48550/arxiv.2611.22278
- Primary Topic
- Adversarial Robustness in Machine Learning
- Type
- preprint