Rui Wu
Publications
Greed Is Learned: Visible Incentives as Reward-Hacking Triggers
Deployed agents increasingly act with their reward proxy in view, such as a balance, score, or KPI dashboard. We show that reinforcement learning can make a policy \emph{addicted} to such a visible self-benefit channel. It chases the displayed payoff across held-out domains, sacrifices the true task to do so, and follows the channel wherever we rewrite it, while policies that never saw the channel stay honest. We call this \emph{reward-channel addiction} and study it in \emph{MoneyWorld}, a synthetic sandbox. The addiction can \emph{flip a model's safety alignment}: trained only on innocuous money tasks with no safety content, the model abandons the safe action it otherwise always takes whenever a dashboard pays for an unsafe one, and reverts to safe once the channel is hidden. This learned bribe replicates across model scales and families. Blindly optimizing super-capable, next-generation AI on KPIs or P\&L can be dangerous for alignment. \emph{Greed is learned} when following such a channel pays.
Taming Actor-Observer Asymmetry in Agents via Dialectical Alignment
Large Language Model agents have rapidly evolved from static text generators into dynamic systems capable of executing complex autonomous workflows. To enhance reliability, multi-agent frameworks assigning specialized roles are increasingly adopted to enable self-reflection and mutual auditing. While such role-playing effectively leverages domain expert knowledge, we find it simultaneously induces a human-like cognitive bias known as Actor-Observer Asymmetry (AOA). Specifically, an agent acting as an actor (during self-reflection) tends to attribute failures to external factors, whereas an observer (during mutual auditing) attributes the same errors to internal faults. We quantify this using our new Ambiguous Failure Benchmark, which reveals that simply swapping perspectives triggers the AOA effect in over 20% of cases for most models. To tame this bias, we introduce ReTAS (Reasoning via Thesis-Antithesis-Synthesis), a model trained through dialectical alignment to enforce perspective-invariant reasoning. By integrating dialectical chain-of-thought with Group Relative Policy Optimization, ReTAS guides agents to synthesize conflicting viewpoints into an objective consensus. Experiments demonstrate that ReTAS effectively mitigates attribution inconsistency and significantly improves fault resolution rates in ambiguous scenarios.
Chinese Labor Law Large Language Model Benchmark
Recent advances in large language models (LLMs) have led to substantial progress in domain-specific applications, particularly within the legal domain. However, general-purpose models such as GPT-4 often struggle with specialized subdomains that require precise legal knowledge, complex reasoning, and contextual sensitivity. To address these limitations, we present LabourLawLLM, a legal large language model tailored to Chinese labor law. We also introduce LabourLawBench, a comprehensive benchmark covering diverse labor-law tasks, including legal provision citation, knowledge-based question answering, case classification, compensation computation, named entity recognition, and legal case analysis. Our evaluation framework combines objective metrics (e.g., ROUGE-L, accuracy, F1, and soft-F1) with subjective assessment based on GPT-4 scoring. Experiments show that LabourLawLLM consistently outperforms general-purpose and existing legal-specific LLMs across task categories. Beyond labor law, our methodology provides a scalable approach for building specialized LLMs in other legal subfields, improving accuracy, reliability, and societal value of legal AI applications.
Reasoning over Precedents Alongside Statutes: Case-Augmented Deliberative Alignment for LLM Safety
Ensuring that Large Language Models (LLMs) adhere to safety principles without refusing benign requests remains a significant challenge. While OpenAI introduces deliberative alignment (DA) to enhance the safety of its o-series models through reasoning over detailed ``code-like'' safety rules, the effectiveness of this approach in open-source LLMs, which typically lack advanced reasoning capabilities, is understudied. In this work, we systematically evaluate the impact of explicitly specifying extensive safety codes versus demonstrating them through illustrative cases. We find that referencing explicit codes inconsistently improves harmlessness and systematically degrades helpfulness, whereas training on case-augmented simple codes yields more robust and generalized safety behaviors. By guiding LLMs with case-augmented reasoning instead of extensive code-like safety rules, we avoid rigid adherence to narrowly enumerated rules and enable broader adaptability. Building on these insights, we propose CADA, a case-augmented deliberative alignment method for LLMs utilizing reinforcement learning on self-generated safety reasoning chains. CADA effectively enhances harmlessness, improves robustness against attacks, and reduces over-refusal while preserving utility across diverse benchmarks, offering a practical alternative to rule-only DA for improving safety while maintaining helpfulness.