cs.CRNov 24, 2025

From Strategy Search to Policy Learning: MCTS-Guided Adversarial Co-Evolution for LLM Safety Alignment

Authors: Xurui Li, Kaisong Song

Organizations: Fudan University, China · Alibaba Group, China

Abstract

Large Language Models (LLMs) increasingly power web-facing applications, including conversational assistants and content services, where open user interaction exposes them to diverse safety threats and adaptive attacks. In particular, jailbreak attackers repeatedly reformulate harmful requests to bypass safety alignment. Although existing studies have concentrated on defending dynamic jailbreaks, they often retain discovered attacks as flat samples and discard information about how those attacks were found. In this paper, we introduce MACE, an MCTS-guided adversarial co-evolution framework that turns search into policy-learning signals for both an attacker and a defender. Its search component, GS-MCTS, explores multi-step compositions of rewriting operators and associates discovered attacks with empirical success frequency and path depth. Its optimization component, AC-TGPO, uses these statistics for difficulty-aware group-relative credit assignment and constructs a temporally asymmetric curriculum from current interactions, pairings between the latest attacker and preceding defender, and attacks that survive a defense update. Thus, MACE connects search structure to both within-round policy optimization and cross-round memory, rather than using MCTS only to generate attacks or treating each round as a flat adversarial dataset. Removing difficulty-aware weighting, asymmetric samples, or persistent hard samples increases defense ASR-LR by 3.7 %, 6.0 %, and 5.0 %, respectively. Across evaluated models, the resulting attacker achieves the highest ASR-LR in all six model--dataset settings, while the defender reduces average ASR-LR from 11.9 % for the best baseline to 8.8 %.

Figures & tables

Explore similar work

Nov 17, 2026

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: attack strategy collapse, where the attacker overfits to a narrow set of high-reward rewrites, and 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 \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 normalized marginal coverage gain reward, providing a bounded, non-vanishing exploration signal at the strategy layer. On the defender side, \method maintains a unified 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.
May 11, 2026

Metis: Learning to Jailbreak LLMs via Self-Evolving Metacognitive Policy Optimization

Red teaming is critical for uncovering vulnerabilities in Large Language Models (LLMs). While automated methods have improved scalability, existing approaches often rely on static heuristics or stochastic search, rendering them brittle against advanced safety alignment. To address this, we introduce Metis, a framework that reformulates jailbreaking as inference-time policy optimization within an adversarial Partially Observable Markov Decision Process (POMDP). Metis employs a self-evolving metacognitive loop to perform causal diagnosis of a target's defense logic and leverages structured feedback as a semantic gradient to refine its policy, offering enhanced interpretability through transparent reasoning traces. Extensive evaluations across 10 diverse models demonstrate that Metis achieves the strongest average Attack Success Rate (ASR) among compared methods at 89.2%, maintaining high efficacy on resilient frontier models (e.g., 76.0% on O1 and 78.0% on GPT-5-chat) where traditional baselines exhibit substantial performance degradation. By replacing redundant exploration with directed optimization, Metis reduces token costs by an average of 8.2x and up to 11.4x. Our analysis reveals that current defenses remain vulnerable to internally-steered, closed-loop reasoning trajectories under the tested settings, highlighting a critical need for next-generation defenses capable of reasoning about safety dynamically during inference.
May 13, 2026

Model-Agnostic Lifelong LLM Safety via Externalized Attack-Defense Co-Evolution

Large language models remain vulnerable to adversarial prompts that elicit harmful outputs. Existing safety paradigms typically couple red-teaming and post-training in a closed, policy-centric loop, causing attack discovery to suffer from rapid saturation and limiting the exposure of novel failure modes, while leaving defenses inefficient, rigid, and difficult to transfer across victim models. To this end, we propose EvoSafety, an LLM safety framework built around persistent, inspectable, and reusable external structures. For red teaming, EvoSafety equips the attack policy with an adversarial skill library, enabling continued vulnerability probing through simple library expansion after saturation, while supporting the evolution of adversarial vectors. For defense learning, EvoSafety replaces model-specific safety fine-tuning with a lightweight auxiliary defense model augmented with memory retrieval. This enables efficient, transferable, and model-agnostic safety improvements, while allowing robustness to be enhanced solely through memory updates. With a single training procedure, the defense policy can operate in both Steer and Guard modes: the former activates the victim model's intrinsic defense mechanisms, while the latter directly filters harmful inputs. Extensive experiments demonstrate the superiority of EvoSafety: in Guard mode, it achieves a 99.61% defense success rate, outperforming Qwen3Guard-8B by 14.13% with only 37.5% of its parameters, while preserving reasoning performance on benign queries. Warning: This paper contains potentially harmful text.