English
Related papers

Related papers: Policy-Invisible Violations in LLM-Based Agents

200 papers

As LLM-based agents operate over sequential multi-step reasoning, hallucinations arising at intermediate steps risk propagating along the trajectory, thus degrading overall reliability. Unlike hallucination detection in single-turn…

Computation and Language · Computer Science 2026-01-13 Xuannan Liu , Xiao Yang , Zekun Li , Peipei Li , Ran He

Autonomous agents have rapidly matured as task executors and seen widespread deployment via harnesses such as OpenClaw. Safety concerns have rightly drawn growing research attention, and beneath them lie the values silently steering agent…

Artificial Intelligence · Computer Science 2026-05-12 Haonan Dong , Qiguan Feng , Kehan Jiang , Haoran Ye , Xin Zhang , Guojie Song

Large Language Model (LLM) agents hold promise for a flexible and scalable alternative to traditional business process automation, but struggle to reliably follow complex company policies. In this study we introduce a deterministic,…

Computation and Language · Computer Science 2025-10-07 Naama Zwerdling , David Boaz , Ella Rabinovich , Guy Uziel , David Amid , Ateret Anaby-Tavor

Tool-using large language model (LLM) agents are increasingly deployed in settings where their reliable behavior is governed by strict procedural manuals. Ensuring that such agents comply with the rules from these manuals is challenging, as…

Computation and Language · Computer Science 2026-05-08 Ashwani Anand , Ivi Chatzi , Ritam Raha , Anne-Kathrin Schmuck

Large language models (LLMs) are increasingly used as simulated participants in social science experiments, but their behavior is often unstable and highly sensitive to design choices. Prior evaluations frequently conflate base-model…

Artificial Intelligence · Computer Science 2026-02-03 Xuan Liu , Haoyang Shang , Zizhang Liu , Xinyan Liu , Yunze Xiao , Yiwen Tu , Haojian Jin

LLM agents operating under organizational policies must comply with authorization constraints typically specified in natural language. In practice, such specifications inevitably contain ambiguities and logical or semantic gaps that cause…

Computation and Language · Computer Science 2026-04-20 Jihye Choi , Jinsung Yoon , Long T. Le , Somesh Jha , Tomas Pfister

While Large Language Models (LLMs) have evolved into tool-using agents, they remain brittle in long-horizon interactions. Unlike mathematical reasoning where errors are often rectifiable via backtracking, tool-use failures frequently induce…

Artificial Intelligence · Computer Science 2026-03-17 Shengda Fan , Xuyan Ye , Yupeng Huo , Zhi-Yuan Chen , Yiju Guo , Shenzhi Yang , Wenkai Yang , Shuqi Ye , Jingwen Chen , Haotian Chen , Xin Cong , Yankai Lin

AI agents are increasingly embedded in real software systems, where they execute multi-step workflows through multi-turn dialogue, tool invocations, and intermediate decisions. These long execution histories, called agentic traces, make…

Software Engineering · Computer Science 2026-05-12 Reshabh K Sharma , Shraddha Barke , Benjamin Zorn

Large vision-language model (LVLM)-based web agents are emerging as powerful tools for automating complex online tasks. However, when deployed in real-world environments, they face serious security risks, motivating the design of security…

Cryptography and Security · Computer Science 2026-04-15 Zonghao Ying , Yangguang Shao , Jianle Gan , Gan Xu , Wenxin Zhang , Quanchen Zou , Junzheng Shi , Zhenfei Yin , Mingchuan Zhang , Aishan Liu , Xianglong Liu

Large language models in regulated financial workflows are governed by natural-language policies that the same model interprets, creating a principal--agent failure: outputs can appear compliant without being compliant. Existing evaluation…

Computation and Language · Computer Science 2026-05-15 José Manuel de la Chica Rodríguez , Carlos Martí-González

LLM-driven GUI agents are increasingly used in production systems to automate workflows and simulate users for evaluation and optimization. Yet most GUI-agent evaluations emphasize task success and provide limited evidence on whether agents…

Information Retrieval · Computer Science 2026-04-10 Maria Movin , Claudia Hauff , Aron Henriksson , Panagiotis Papapetrou

Individuals' concerns about data privacy and AI safety are highly contextualized and extend beyond sensitive patterns. Addressing these issues requires reasoning about the context to identify and mitigate potential risks. Though researchers…

Computation and Language · Computer Science 2026-04-15 Haoran Li , Yulin Chen , Huihao Jing , Wenbin Hu , Tsz Ho Li , Chanhou Lou , Hong Ting Tsang , Sirui Han , Yangqiu Song

When Agent A delegates to Agent B, which invokes Tool C on behalf of User X, no existing framework can answer: whose authorization chain led to this action, and where did it violate policy? This paper introduces SentinelAgent, a formal…

Cryptography and Security · Computer Science 2026-04-06 KrishnaSaiReddy Patil

Verifying LLM-generated systems code is hard: bugs are prevalent, formal specifications are missing, and safety contracts are encoded implicitly at call sites rather than enforced at function boundaries. We propose agentic model checking, a…

Software Engineering · Computer Science 2026-05-21 Youcheng Sun , Jiawen Liu , Daniel Kroening , Jason Xue

LLM-as-a-Judge pipelines have become the de facto evaluator for agent safety, yet existing benchmarks treat their verdicts as ground-truth proxies without checking whether the verdicts depend on the agent's behavior or merely on how the…

Artificial Intelligence · Computer Science 2026-05-08 Shihao Weng , Yang Feng , Xiaofei Xie

Multi-agent systems powered by large language models (LLMs) are transforming enterprise automation, yet systematic evaluation methodologies for assessing tool-use reliability remain underdeveloped. We introduce a comprehensive diagnostic…

Artificial Intelligence · Computer Science 2026-01-26 Donghao Huang , Gauri Malwe , Zhaoxia Wang

As LLM-based agents increasingly operate in high-stakes domains with real-world consequences, ensuring their behavioral safety becomes paramount. The dominant oversight paradigm, LLM-as-a-Judge, faces a fundamental dilemma: how can…

Artificial Intelligence · Computer Science 2026-02-13 Jiayi Zhou , Yang Sheng , Hantao Lou , Yaodong Yang , Jie Fu

When an LLM agent reads a confidential file, then writes a summary, then emails it externally, no single step is unsafe, but the sequence is a data leak. We call this safety drift: individually safe actions compounding into violations.…

Cryptography and Security · Computer Science 2026-03-31 Aditya Dhodapkar , Farhaan Pishori

LLM-based agents have demonstrated promising adaptability in real-world applications. However, these agents remain vulnerable to a wide range of attacks, such as tool poisoning and malicious instructions, that compromise their execution…

Cryptography and Security · Computer Science 2025-10-14 Jiahao Liu , Bonan Ruan , Xianglin Yang , Zhiwei Lin , Yan Liu , Yang Wang , Tao Wei , Zhenkai Liang

Enterprise LLM agents can dramatically improve workplace productivity, but their core capability, retrieving and using internal context to act on a user's behalf, also creates new risks for sensitive information leakage. We introduce…

Cryptography and Security · Computer Science 2026-04-24 Wenjie Fu , Xiaoting Qin , Jue Zhang , Qingwei Lin , Lukas Wutschitz , Robert Sim , Saravan Rajmohan , Dongmei Zhang