# The Law of Knowledge Overshadowing: Towards Understanding, Predicting, and Preventing LLM Hallucination

arXiv: https://arxiv.org/abs/2502.16143 · HTML: https://mlsanity.com/paper/2502.16143/

## Key facts

- Cited by 23 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 22 Feb 2025.

## Built on by (most cited first, 23 of 23)

- [KnowRL: Exploring Knowledgeable Reinforcement Learning for Factuality](https://mlsanity.com/paper/2506.19807/) (2025)
- [Graph4MM: Weaving Multimodal Learning with Structural Information](https://mlsanity.com/paper/2510.16990/) (2025)
- [GeoEdit: Geometric Knowledge Editing for Large Language Models](https://mlsanity.com/paper/2502.19953/) (2025)
- [Do Role-Playing Agents Practice What They Preach? Belief-Behavior Consistency in LLM-Based Simulations of Human Trust](https://mlsanity.com/paper/2507.02197/) (2025)
- [CompliBench: Benchmarking LLM Judges for Compliance Violation Detection in Dialogue Systems](https://mlsanity.com/paper/2604.12312/) (2026)
- [ModelingAgent: Bridging LLMs and Mathematical Modeling for Real-World Challenges](https://mlsanity.com/paper/2505.15068/) (2025)
- [Long-Tail Knowledge in Large Language Models: Taxonomy, Mechanisms, Interventions and Implications](https://mlsanity.com/paper/2602.16201/) (2026)
- [HalluSAE: Detecting Hallucinations in Large Language Models via Sparse Auto-Encoders](https://mlsanity.com/paper/2604.16430/) (2026)
- [RePPL: Recalibrating Perplexity by Uncertainty in Semantic Propagation and Language Generation for Explainable QA Hallucination Detection](https://mlsanity.com/paper/2505.15386/) (2025)
- [Geometric-disentangelment Unlearning](https://mlsanity.com/paper/2511.17100/) (2025)
- [Revisiting Hallucination Detection with Effective Rank-based Uncertainty](https://mlsanity.com/paper/2510.08389/) (2025)
- [Concept Incongruence: An Exploration of Time and Death in Role Playing](https://mlsanity.com/paper/2505.14905/) (2025)
- [How Significant Are the Real Performance Gains? An Unbiased Evaluation Framework for GraphRAG](https://mlsanity.com/paper/2506.06331/) (2025)
- [Analyzing and Internalizing Complex Policy Documents for LLM Agents](https://mlsanity.com/paper/2510.11588/) (2025)
- [Large Language Models and Scientific Discourse: Where's the Intelligence?](https://mlsanity.com/paper/2603.23543/) (2026)
- [Predictable Confabulations: Factual Recall by LLMs Scales with Model Size and Topic Frequency](https://mlsanity.com/paper/2605.18732/) (2026)
- [ActiShade: Activating Overshadowed Knowledge to Guide Multi-Hop Reasoning in Large Language Models](https://mlsanity.com/paper/2601.07260/) (2026)
- [Attribution Techniques for Mitigating Hallucinated Information in RAG Systems: A Survey](https://mlsanity.com/paper/2601.19927/) (2026)
- [Estimating the Black-box LLM Uncertainty with Distribution-Aligned Adversarial Distillation](https://mlsanity.com/paper/2605.05777/) (2026)
- [Scalable Token-Level Hallucination Detection in Large Language Models](https://mlsanity.com/paper/2605.12384/) (2026)
- [Does Finetuning with Scientific Data Increase Hallucinations? A Multi-domain Factuality Evaluation of LLMs](https://mlsanity.com/paper/2606.21359/) (2026)
- [Reliability Scales Inversely: Hallucinations Snowball Faster in Bigger Language Models](https://mlsanity.com/paper/2607.18292/) (2026)
- [Right Number, Wrong State? Measuring Cross-Jurisdiction Substitution in LLM Recall of State Policy](https://mlsanity.com/paper/2610.09458/) (2026)

## Abstract (arXiv)

Hallucination is a persistent challenge in large language models (LLMs), where even with rigorous quality control, models often generate distorted facts. This paradox, in which error generation continues despite high-quality training data, calls for a deeper understanding of the underlying LLM mechanisms. To address it, we propose a novel concept: knowledge overshadowing, where model's dominant knowledge can obscure less prominent knowledge during text generation, causing the model to fabricate inaccurate details. Building on this idea, we introduce a novel framework to quantify factual hallucinations by modeling knowledge overshadowing. Central to our approach is the log-linear law, which predicts that the rate of factual hallucination increases linearly with the logarithmic scale of (1) Knowledge Popularity, (2) Knowledge Length, and (3) Model Size. The law provides a means to preemptively quantify hallucinations, offering foresight into their occurrence even before model training or inference. Built on overshadowing effect, we propose a new decoding strategy CoDa, to mitigate hallucinations, which notably enhance model factuality on Overshadow (27.9%), MemoTrap (13.1%) and NQ-Swap (18.3%). Our findings not only deepen understandings of the underlying mechanisms behind hallucinations but also provide actionable insights for developing more predictable and controllable language models.