The paper introduces Global PSRO, a novel deep reinforcement learning framework that efficiently approximates Nash equilibria in large two-player zero-sum games by intelligently expanding the strategy…

S-SPPO introduces a dual-space semantic calibration framework to stabilize Self-Play Preference Optimization (SPPO), preventing policy degeneration when preference oracles assign overly confident wins…

The paper proposes DNQ, a scalable solver-in-the-loop framework for training agents in multi-turn simultaneous bidding games by leveraging pairwise payoff estimation to approximate complex equilibrium…

The paper introduces an outer-loop AI agent that autonomously redesigns LLM policy-synthesis pipelines for multi-agent social dilemmas, demonstrating that the optimal pipeline structure depends critic…

SafeMCP is a server-side defense plugin that uses look-ahead reasoning to proactively filter and constrain tool acquisition for LLM agents, thereby mitigating catastrophic risks associated with expand…

The paper proposes a feasible-reward-set framework to perform Inverse Reinforcement Learning (IRL) when data comes from multiple imperfect demonstrators, providing theoretical guarantees and practical…

The paper introduces RoleCDE, a novel benchmark that evaluates role-playing agents' ability to resolve conflicts between role-specific values and general alignment constraints, revealing a 'Role Value…

ReSkill is an RL-in-the-loop framework that reconciles skill creation and policy optimization by automatically creating, testing, and refining modular skills alongside the agent's policy learning, lea…

The paper introduces Prompted Policy Optimization (PromptPO), an LLM-based method that successfully optimizes policies for various sequential RL tasks, demonstrating that LLMs can replace classical RL…

The paper proposes S2L-PO, a framework that uses smaller, naturally diverse models as structured explorers to enhance the policy-level diversity and performance of larger language models during traini…

The paper introduces a novel shielding framework for Robust MDPs (RMDPs) that guarantees safety under worst-case transition probabilities, enabling safe reinforcement learning even when transition dyn…

ARES is a novel framework that systematically discovers and mitigates dual vulnerabilities in RLHF systems by simultaneously testing the core LLM and its Reward Model (RM) using structured adversarial…

This paper investigates how on-policy Reinforcement Learning (RL) affects LLM safety, finding that safety training modulates harmful misalignment, but the direction of this effect is highly dependent…

The paper proposes a novel, explicitly exploratory iterative Nash Learning from Human Feedback (NLHF) algorithm that achieves strong regret bounds for optimizing LLMs based on complex, non-scalar huma…

The EvoSafety framework enhances LLM safety by externalizing attack and defense mechanisms, enabling persistent, transferable, and model-agnostic robustness against adversarial prompts.

CARE-RL introduces a framework combining protocol-aware reward generation and capability-aware optimization to effectively mitigate cross-domain conflicts in multi-domain reinforcement learning for LL…

The paper proposes a mutagenic incentive intervention approach that mitigates collusion in embodied multi-agent systems by reshaping agents' payoff structures, effectively inducing defection and maint…

The paper introduces RUBAS, a rubric-based reinforcement learning framework that improves agent safety by providing fine-grained, multi-dimensional rewards for complex tool-use scenarios.

The paper proposes Hysteretic Policy Optimization (HPO) and its adaptive variant (A-HPO) to stabilize reinforcement learning training in sparse-reward environments by better balancing positive and neg…

The paper introduces Mindgames, a comprehensive multi-game arena for evaluating LLM agents' sustained social and strategic reasoning, demonstrating that current evaluations are limited by structural s…