Is Your LLM-as-a-Recommender Agent Trustable? LLMs' Recommendation is Easily Hacked by Biases (Preferences)

• URL: http://arxiv.org/abs/2603.17417v2
• Date: Fri, 20 Mar 2026 13:11:53 GMT
• Title: Is Your LLM-as-a-Recommender Agent Trustable? LLMs' Recommendation is Easily Hacked by Biases (Preferences)
• Abstract summary: We introduce a textbfBias textbfRecommendation textbfBenchmark to highlight the critical vulnerability of such agents to biases in high-value real-world tasks.<n>The benchmark includes three practical domains: paper review, e-commerce, and job recruitment.
• Score: 28.6608041553667
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Current Large Language Models (LLMs) are gradually exploited in practically valuable agentic workflows such as Deep Research, E-commerce recommendation, and job recruitment. In these applications, LLMs need to select some optimal solutions from massive candidates, which we term as \textit{LLM-as-a-Recommender} paradigm. However, the reliability of using LLM agents for recommendations is underexplored. In this work, we introduce a \textbf{Bias} \textbf{Rec}ommendation \textbf{Bench}mark (\textbf{BiasRecBench}) to highlight the critical vulnerability of such agents to biases in high-value real-world tasks. The benchmark includes three practical domains: paper review, e-commerce, and job recruitment. We construct a \textsc{Bias Synthesis Pipeline with Calibrated Quality Margins} that 1) synthesizes evaluation data by controlling the quality gap between optimal and sub-optimal options to provide a calibrated testbed to elicit the vulnerability to biases; 2) injects contextual biases that are logical and suitable for option contexts. Extensive experiments on both SOTA (Gemini-{2.5,3}-pro, GPT-4o, DeepSeek-R1) and small-scale LLMs reveal that agents frequently succumb to injected biases despite having sufficient reasoning capabilities to identify the ground truth. These findings expose a significant reliability bottleneck in current agentic workflows, calling for specialized alignment strategies for LLM-as-a-Recommender. The complete code and evaluation datasets will be made publicly available shortly.

Related papers

• LLM-as-a-Verifier: A General-Purpose Verification Framework [74.40111651545979]
  We introduce LLM-as-a-Verifier, a general-purpose verification framework.<n>It provides fine-grained feedback for agentic tasks without requiring additional training.<n>It achieves state-of-the-art performance on several benchmarks.
  arXiv  Detail & Related papers  (2026-07-06T17:59:35Z)
• Towards Position-Robust Talent Recommendation via Large Language Models [2.145037859243627]
  We propose L3TR, a novel framework for listwise talent recommendation with large language models (LLMs)<n>In this paper, we propose a block attention mechanism and a local positional encoding method to enhance inter-document processing and mitigate the position bias and concurrent token bias issue.<n>Experiments on two real-world datasets validated the effectiveness of L3TR, showing consistent improvements over existing baselines.
  arXiv  Detail & Related papers  (2026-04-02T15:54:03Z)
• RAVEL: Reasoning Agents for Validating and Evaluating LLM Text Synthesis [78.32151470154422]
  We introduce RAVEL, an agentic framework that enables the testers to autonomously plan and execute typical synthesis operations.<n>We present C3EBench, a benchmark comprising 1,258 samples derived from professional human writings.<n>By augmenting RAVEL with SOTA LLMs as operators, we find that such agentic text synthesis is dominated by the LLM's reasoning capability.
  arXiv  Detail & Related papers  (2026-02-28T14:47:34Z)
• Evaluating Position Bias in Large Language Model Recommendations [3.430780143519032]
  Large Language Models (LLMs) are being increasingly explored as general-purpose tools for recommendation tasks.<n>We show that LLM-based recommendation models suffer from position bias, where the order of candidate items in a prompt can disproportionately influence the recommendations produced by LLMs.<n>We introduce a new prompting strategy to mitigate the position bias of LLM recommendation models called Ranking via Iterative SElection.
  arXiv  Detail & Related papers  (2025-08-04T03:30:26Z)
• Towards Next-Generation Recommender Systems: A Benchmark for Personalized Recommendation Assistant with LLMs [38.83854553636802]
  Large language models (LLMs) have revolutionized the foundational architecture of RecSys.<n>Most existing studies rely on fixed task-specific prompt templates to generate recommendations.<n>This is because commonly used datasets lack high-quality textual user queries that reflect real-world recommendation scenarios.<n>We introduce RecBench+, a new dataset benchmark designed to access LLMs' ability to handle intricate user recommendation needs.
  arXiv  Detail & Related papers  (2025-03-12T13:28:23Z)
• The Dual-use Dilemma in LLMs: Do Empowering Ethical Capacities Make a Degraded Utility? [54.18519360412294]
  Large Language Models (LLMs) must balance between rejecting harmful requests for safety and accommodating legitimate ones for utility.<n>This paper presents a Direct Preference Optimization (DPO) based alignment framework that achieves better overall performance.<n>We analyze experimental results obtained from testing DeepSeek-R1 on our benchmark and reveal the critical ethical concerns raised by this highly acclaimed model.
  arXiv  Detail & Related papers  (2025-01-20T06:35:01Z)
• LLM-Assisted Relevance Assessments: When Should We Ask LLMs for Help? [20.998805709422292]
  Test collections are information-retrieval tools that allow researchers to quickly and easily evaluate ranking algorithms.<n>As a cheaper alternative, recent studies have proposed using large language models (LLMs) to completely replace human assessors.<n>We propose LARA, an effective method to balance manual annotations with LLM annotations, helping build a rich and reliable test collection even under a low budget.
  arXiv  Detail & Related papers  (2024-11-11T11:17:35Z)
• LLM Self-Correction with DeCRIM: Decompose, Critique, and Refine for Enhanced Following of Instructions with Multiple Constraints [86.59857711385833]
  We introduce RealInstruct, the first benchmark designed to evaluate LLMs' ability to follow real-world multi-constrained instructions. To address the performance gap between open-source and proprietary models, we propose the Decompose, Critique and Refine (DeCRIM) self-correction pipeline. Our results show that DeCRIM improves Mistral's performance by 7.3% on RealInstruct and 8.0% on IFEval even with weak feedback.
  arXiv  Detail & Related papers  (2024-10-09T01:25:10Z)
• LLM4MSR: An LLM-Enhanced Paradigm for Multi-Scenario Recommendation [52.55639178180821]
  The study on multi-scenario recommendation (MSR) has attracted much attention, which uses the data from all scenarios to simultaneously improve their recommendation performance.<n>Existing methods tend to integrate insufficient scenario knowledge and neglect learning personalized cross-scenario preferences, thus leading to sub-optimal performance.<n>We propose a large language model (LLM)-enhanced paradigm LLM4MSR to fill these gaps.
  arXiv  Detail & Related papers  (2024-06-18T11:59:36Z)
• Self-Supervised Position Debiasing for Large Language Models [39.261233221850155]
  We propose a self-supervised position debiasing (SOD) framework to mitigate position bias for large language models (LLMs) Experiments on eight datasets and five tasks show that SOD consistently outperforms existing methods in mitigating three types of position biases.
  arXiv  Detail & Related papers  (2024-01-02T14:12:41Z)
• Large Language Models are Not Stable Recommender Systems [45.941176155464824]
  We introduce exploratory research and find consistent patterns of positional bias in large language models (LLMs) We propose a Bayesian probabilistic framework, STELLA (Stable LLM for Recommendation), which involves a two-stage pipeline. Our framework can capitalize on existing pattern information to calibrate instability of LLMs, and enhance recommendation performance.
  arXiv  Detail & Related papers  (2023-12-25T14:54:33Z)
• LLMRefine: Pinpointing and Refining Large Language Models via Fine-Grained Actionable Feedback [65.84061725174269]
  Recent large language models (LLM) are leveraging human feedback to improve their generation quality. We propose LLMRefine, an inference time optimization method to refine LLM's output. We conduct experiments on three text generation tasks, including machine translation, long-form question answering (QA), and topical summarization. LLMRefine consistently outperforms all baseline approaches, achieving improvements up to 1.7 MetricX points on translation tasks, 8.1 ROUGE-L on ASQA, 2.2 ROUGE-L on topical summarization.
  arXiv  Detail & Related papers  (2023-11-15T19:52:11Z)
• Fake Alignment: Are LLMs Really Aligned Well? [91.26543768665778]
  This study investigates the substantial discrepancy in performance between multiple-choice questions and open-ended questions. Inspired by research on jailbreak attack patterns, we argue this is caused by mismatched generalization.
  arXiv  Detail & Related papers  (2023-11-10T08:01:23Z)
• Do-Not-Answer: A Dataset for Evaluating Safeguards in LLMs [59.596335292426105]
  This paper collects the first open-source dataset to evaluate safeguards in large language models. We train several BERT-like classifiers to achieve results comparable with GPT-4 on automatic safety evaluation.
  arXiv  Detail & Related papers  (2023-08-25T14:02:12Z)
• Is ChatGPT Fair for Recommendation? Evaluating Fairness in Large Language Model Recommendation [52.62492168507781]
  We propose a novel benchmark called Fairness of Recommendation via LLM (FaiRLLM) This benchmark comprises carefully crafted metrics and a dataset that accounts for eight sensitive attributes. By utilizing our FaiRLLM benchmark, we conducted an evaluation of ChatGPT and discovered that it still exhibits unfairness to some sensitive attributes when generating recommendations.
  arXiv  Detail & Related papers  (2023-05-12T16:54:36Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.