7 papers
Safety, or Just Capability? A Validity Audit of Agent-Safety Benchmarks
Youting Wang, Xiao Han, Dingyan Shang +2
Agent-safety benchmarks measure different behaviors, and their scores get quoted interchangeably as an agent's safety. We treat four of them (R-Judge, InjecAgent, AgentHarm, AgentD…
Context-Masked Truncated Reasoning Audits for Answer-Key Dependence in LLM Tutors
Bonan Shen, Dingyan Shang, Youting Wang +2
Large language model (LLM) tutors may have access to teacher notes, answer keys, rubrics, or retrieved solutions while producing student-facing explanations. We study whether trunc…
Self-Commitment Latency: A Reward-Free Probe for Prompted Implicit Hacking
Bonan Shen, Youting Wang, Dingyan Shang +1
Implicit reward hacking is hard to audit when a language model's chain of thought appears benign: a final answer may be anchored by a prompt shortcut while the written reasoning st…
VisualLeakBench: Reproducible Action-Boundary Propagation Failures in Vision-Language Agents
Youting Wang, Yuan Tang, Yitian Qian +1
Vision-language agents increasingly consume screenshots, documents, and user interfaces before writing to memory, sending messages, or invoking external tools. We study a concrete…
When LLM Reward Design Fails: Diagnostic-Driven Refinement for Sparse Structured RL
Youting Wang, Yuan Tang, Bowen Liu +2
For sparse, structured reinforcement-learning tasks with semantic reward-function interfaces, LLM-generated reward shaping is better framed as debugging than one-shot generation. W…
Names Are All You Need: Effective and Safe Regression Test Selection for Python
You Wang, Michael Pradel, Zhongxin Liu
Regression test selection reduces the cost of regression testing by executing only those tests affected by a code change. Despite extensive study of RTS in statically typed languag…