# Hallucinations in Code Change to Natural Language Generation: Prevalence and Evaluation of Detection Metrics

arXiv: https://arxiv.org/abs/2508.08661 · HTML: https://mlsanity.com/paper/2508.08661/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Builds on 25 of its 48 references that are in the map.
- First posted on arXiv 12 Aug 2025.

## Builds on (most cited first, 25 of 25)

- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Learning Important Features Through Propagating Activation Differences](https://mlsanity.com/paper/1704.02685/) (2017)
- [Reducing conversational agents' overconfidence through linguistic calibration](https://mlsanity.com/paper/2012.14983/) (2020)
- [Large Language Models for Software Engineering: Survey and Open Problems](https://mlsanity.com/paper/2310.03533/) (2023)
- [Jigsaw: Large Language Models meet Program Synthesis](https://mlsanity.com/paper/2112.02969/) (2021)
- [Automating Code Review Activities by Large-Scale Pre-training](https://mlsanity.com/paper/2203.09095/) (2022)
- [Navigating the Grey Area: How Expressions of Uncertainty and Overconfidence Affect Language Models](https://mlsanity.com/paper/2302.13439/) (2023)
- [Inseq: An Interpretability Toolkit for Sequence Generation Models](https://mlsanity.com/paper/2302.13942/) (2023)
- [Detecting and Mitigating Hallucinations in Machine Translation: Model Internal Workings Alone Do Well, Sentence Similarity Even Better](https://mlsanity.com/paper/2212.08597/) (2022)
- [On Early Detection of Hallucinations in Factual Question Answering](https://mlsanity.com/paper/2312.14183/) (2023)
- [Towards Automating Code Review Activities](https://mlsanity.com/paper/2101.02518/) (2021)
- [CodeHalu: Investigating Code Hallucinations in LLMs via Execution-based Verification](https://mlsanity.com/paper/2405.00253/) (2024)
- [What Makes a Good Commit Message?](https://mlsanity.com/paper/2202.02974/) (2022)
- [Cost-Effective Hallucination Detection for LLMs](https://mlsanity.com/paper/2407.21424/) (2024)
- [Improving Automated Code Reviews: Learning from Experience](https://mlsanity.com/paper/2402.03777/) (2024)
- [Too Noisy To Learn: Enhancing Data Quality for Code Review Comment Generation](https://mlsanity.com/paper/2502.02757/) (2025)
- [Rethinking Code Review Workflows with LLM Assistance: An Empirical Study](https://mlsanity.com/paper/2505.16339/) (2025)
- [CommitBench: A Benchmark for Commit Message Generation](https://mlsanity.com/paper/2403.05188/) (2024)
- [Identifying Inaccurate Descriptions in LLM-generated Code Comments via Test Execution](https://mlsanity.com/paper/2406.14836/) (2024)
- [Reducing Hallucinations in Neural Machine Translation with Feature Attribution](https://mlsanity.com/paper/2211.09878/) (2022)
- [ETF: An Entity Tracing Framework for Hallucination Detection in Code Summaries](https://mlsanity.com/paper/2410.14748/) (2024)
- [Deep Learning-based Code Reviews: A Paradigm Shift or a Double-Edged Sword?](https://mlsanity.com/paper/2411.11401/) (2024)
- [Emotional Strain and Frustration in LLM Interactions in Software Engineering](https://mlsanity.com/paper/2504.10050/) (2025)
- [Towards Practical Defect-Focused Automated Code Review](https://mlsanity.com/paper/2505.17928/) (2025)
- [DiffSearch: A Scalable and Precise Search Engine for Code Changes](https://mlsanity.com/paper/2204.02787/) (2022)

## Built on by (most cited first, 4 of 4)

- [HalluJudge: A Reference-Free Hallucination Detection for Context Misalignment in Code Review Automation](https://mlsanity.com/paper/2601.19072/) (2026)
- [Code Change Characteristics and Description Alignment: A Comparative Study of Agentic versus Human Pull Requests](https://mlsanity.com/paper/2601.17627/) (2026)
- [Is Agentic Code Review Helpful? Mining Developers' Feedback to CodeRabbit Reviews in the Wild](https://mlsanity.com/paper/2607.03316/) (2026)
- [Empirical Analysis and Detection of Hallucinations in LLM-Generated Bug Report Summaries](https://mlsanity.com/paper/2605.24137/) (2026)

## Abstract (arXiv)

Language models have shown strong capabilities across a wide range of tasks in software engineering, such as code generation, yet they suffer from hallucinations. While hallucinations have been studied independently in natural language and code generation, their occurrence in tasks involving code changes which have a structurally complex and context-dependent format of code remains largely unexplored. This paper presents the first comprehensive analysis of hallucinations in two critical tasks involving code change to natural language generation: commit message generation and code review comment generation. We quantify the prevalence of hallucinations in recent language models and explore a range of metric-based approaches to automatically detect them. Our findings reveal that approximately 50% of generated code reviews and 20% of generated commit messages contain hallucinations. Whilst commonly used metrics are weak detectors on their own, combining multiple metrics substantially improves performance. Notably, model confidence and feature attribution metrics effectively contribute to hallucination detection, showing promise for inference-time detection.footnoteAll code and data will be released upon acceptance.