LLM Hallucination Mitigation

LLM: Large Language Model

Overview

The field studies how to make language-model answers more faithful to evidence and more honest about uncertainty. Common approaches include retrieval-augmented generation with claim-level verification, knowledge graphs and structured tools, fine-tuning or reinforcement learning, decoding and activation interventions, and calibrated abstention. Evaluation spans factual question answering, long-form summaries, retrieval-augmented systems, high-stakes medical and legal tasks, vision-language models, and tool-using agents. A central tension is reducing unsupported claims without causing excessive refusal, losing useful information, degrading general capabilities, or adding prohibitive cost and latency.

Latest papers

186 papers9 papers in the last four weeks, down 25% on the four weeks before. 0.1% of all new papers.

Aug 5, 2025

SAVER: Mitigating Hallucinations in Large Vision-Language Models via Style-Aware Visual Early Revision

Large Vision-Language Models (LVLMs) recently achieve significant breakthroughs in understanding complex visual-textual contexts. However, hallucination issues still limit their real-world applicability. Although previous mitigation methods effectively reduce hallucinations in photographic images, they largely overlook the potential risks posed by stylized images, which play crucial roles in critical scenarios such as game scene understanding, art education, and medical analysis. In this work, we first construct a dataset comprising photographic images and their corresponding stylized versions with carefully annotated caption labels. We then conduct head-to-head comparisons on both discriminative and generative tasks by benchmarking 13 advanced LVLMs on the collected datasets. Our findings reveal that stylized images tend to induce significantly more hallucinations than their photographic counterparts. To address this issue, we propose Style-Aware Visual Early Revision SAVER, a novel mechanism that dynamically adjusts LVLMs' final outputs based on the token-level visual attention patterns, leveraging early-layer feedback to mitigate hallucinations caused by stylized images. Extensive experiments demonstrate that SAVER achieves state-of-the-art performance in hallucination mitigation across various models, datasets, and tasks.
Jul 21, 2025

Towards Mitigation of Hallucination for LLM-empowered Agents: Progressive Generalization Bound Exploration and Watchdog Monitor

Empowered by large language models (LLMs), intelligent agents have become a popular paradigm for interacting with open environments to facilitate AI deployment. However, hallucinations generated by LLMs-where outputs are inconsistent with facts-pose a significant challenge, undermining the credibility of intelligent agents. Only if hallucinations can be mitigated, the intelligent agents can be used in real-world without any catastrophic risk. Therefore, effective detection and mitigation of hallucinations are crucial to ensure the dependability of agents. Unfortunately, the related approaches either depend on white-box access to LLMs or fail to accurately identify hallucinations. To address the challenge posed by hallucinations of intelligent agents, we present HalMit, a novel black-box watchdog framework that models the generalization bound of LLM-empowered agents and thus detect hallucinations without requiring internal knowledge of the LLM's architecture. Specifically, a probabilistic fractal sampling technique is proposed to generate a sufficient number of queries to trigger the incredible responses in parallel, efficiently identifying the generalization bound of the target agent. Experimental evaluations demonstrate that HalMit significantly outperforms existing approaches in hallucination monitoring. Its black-box nature and superior performance make HalMit a promising solution for enhancing the dependability of LLM-powered systems.
May 23, 2025

Popular but Wrong: Understanding and Mitigating LLM Overconfidence through Knowledge Popularity

Large language models (LLMs) often produce incorrect answers with high confidence, yet the factors associated with such overconfidence remain insufficiently understood. We study this problem through the lens of knowledge popularity. Using entity-centric factual QA derived from Wikidata triplets, we characterize popularity through question entity popularity, answer popularity, and question-answer co-occurrence. We find two consistent patterns. First, hallucinated answers are far from random: compared with ground-truth answers, they tend to be more popular or more frequently associated with the question entity. Second, confidence is strongly tied to the popularity of generated answers: even among incorrect predictions, more popular answers or those with higher question-answer co-occurrence receive higher confidence. Together, these findings suggest that popular but wrong alternatives may contribute to overconfidence. We further show that popularity-related signals can mitigate overconfidence and improve overall confidence estimation. Across six models and three datasets, incorporating knowledge popularity reduces average confidence on incorrect answers from 0.765 to 0.254 and overall ECE from 0.356 to 0.050, while improving Alignment from 77.08% to 83.72%.
May 14, 2025

Atomic Consistency Preference Optimization for Long-Form Question Answering

Large Language Models (LLMs) often produce factoid hallucinations - plausible yet incorrect answers. A common mitigation strategy is model alignment, which improves factual accuracy by training on curated (factual, non-factual) pairs. However, this approach often relies on a stronger model (e.g., GPT-4) or an external knowledge base to assess factual correctness that may not always be accessible. Addressing this, we propose Atomic Consistency Preference Optimization (ACPO), a self-supervised preference-tuning method that enhances factual accuracy without external supervision. ACPO leverages atomic consistency signals (i.e., the agreement of individual facts across multiple stochastic responses) to identify high- and low-quality data pairs for model alignment. Despite being fully self-supervised, ACPO outperforms the strong supervised alignment baseline by 1.95 points averaged across Phi-3 and Llama3 on the LongFact and BioGen datasets, demonstrating its effectiveness in improving factual reliability without relying on external models or knowledge bases.
Mar 3, 2025

HoT: Highlighted Chain of Thought for Referencing Supporting Facts from Inputs

An Achilles heel of Large Language Models (LLMs) is their tendency to hallucinate non-factual statements. A response mixed of factual and non-factual statements poses a challenge for humans to verify and accurately base their decisions on. To combat this problem, we propose Highlighted Chain-of-Thought Prompting (HoT), a technique for prompting LLMs to generate responses with XML tags that ground facts to those provided in the question. That is, given an input question, LLMs would first re-format the question to add XML tags highlighting key facts, and then, generate a response with highlights over the facts referenced from the input. Compared to vanilla chain of thought prompting (CoT), HoT reduces the rate of hallucination and separately improves LLM accuracy consistently on over 22 tasks from arithmetic, reading comprehension, to logical reasoning. When asking humans to verify LLM responses, highlights help time-limited participants to more accurately and efficiently recognize when LLMs are correct. Yet, surprisingly, when LLMs are wrong, HoTs tend to fool users into believing that an answer is correct.
Jan 19, 2025

Hallucination Mitigation using Agentic AI Natural Language-Based Frameworks

Hallucinations remain a significant challenge for Generative AI, undermining trust in AI systems and their reliability. This study investigates how orchestrating multiple specialized AI agents can mitigate hallucinations, with a focus on systems that use Natural Language Processing (NLP) for agent interaction. We design a pipeline that feeds 310 prompts, purposefully crafted to induce hallucinations, into a front-end agent; its outputs are reviewed and refined by second- and third-level agents that detect unverified claims, add explicit disclaimers and clarify speculative content. We introduce four novel Key Performance Indicators (KPIs) - Factual Claim Density, Factual Grounding References, Fictional Disclaimer Frequency and Explicit Contextualization Score - aggregated into a Total Hallucination Score (THS) and assessed by a dedicated fourth-level evaluator agent. Inter-agent context is carried by the Open Floor Protocol (OFP, formerly OVON, Open Voice Network), whose structured JSON messages let each reviewer pass its assessment of hallucination likelihood, and the reasons behind it, to the next stage. Across the 310 prompts, THS improves by a mean of -0.131 per prompt from the front-end agent to the final reviewer (Cohen's d = 2.38), with lower scores on 302 of 310 prompts. The gain is carried mainly by explicit disclaimers and contextualization, making speculative content more transparent and clearly demarcated from factual claims. The results show that multi-agent orchestration with structured, NLP-based exchange of meta-information can substantially strengthen the epistemic framing of AI-generated responses and improve their explainability.