Elamur Research⌕
☰
home / papers / 2026 / HAD: HAllucination Detection Language Models Based on a Comprehensive Hallucination Taxonomy

HAD: HAllucination Detection Language Models Based on a Comprehensive Hallucination Taxonomy

Paper · 2026 · Fan Xu, Xinyu Hu, Zhenghan Yu et al.

Summary

Large language models often produce fluent text that is wrong or that drifts from the given instruction or input. Existing detectors usually cover only factual errors or only faithfulness errors, target a single task, or use coarse labels. The authors propose a three-level taxonomy with 11 fine-grained hallucination types. It covers faithfulness problems (instruction, input-context and internal inconsistencies) and factuality problems (fact contradiction and fact fabrication).

Guided by this taxonomy, they used GPT-4o to inject hallucinations into correct outputs from many NLG tasks, then filtered the results automatically. This produced roughly 90K training samples. They fine-tuned Qwen2.5 models on this data to obtain HAD, a model that, in one generation pass, names the hallucination type, marks the faulty span and rewrites the output. They also built HADTest, a manually annotated test set.

HAD generally beats both general-purpose LLMs and specialised detectors, reaching state-of-the-art results on HaluEval, FactCHD and FaithBench. This suggests that one model can check hallucinations across many tasks.

Key points

  • The taxonomy has 11 types, including Task Type Inconsistency, Baseless Information, Information Omission, Factual Recall Error, Fabricated Entity and Fictional Attribution.
  • Training data comes from ELI5, Super-NaturalInstruction, GSM8K, FaithDial and Alpaca. GPT-4o (gpt-4o-2024-08-06) injected the hallucinations, and 82.3% of candidates passed filtering, leaving 90,088 hallucinated samples.
  • HAD models are fine-tuned from Qwen2.5-7B-Instruct and Qwen2.5-14B-Instruct.
  • HADTest contains 2,248 human-annotated samples, half hallucinated and half correct.
  • A key limitation is that HAD detects only one hallucinated span per output, and its training relies heavily on synthetic rather than naturally occurring errors.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓Peking UniversityBeijing · China244 works · 39,359 citations🎓Fudan UniversityShanghai · China167 works · 20,987 citations🏢Alibaba GroupHangzhou · China168 works · 15,386 citations

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksComputer Vision and Pattern Recognition · 2,869 works

Cite

Cite as Fan Xu, Xinyu Hu, Zhenghan Yu, Li Lin, Xu Zhang, Yang Zhang, Wei Zhou, Jinjie Gu and Xiaojun Wan. 2026. HAD: HAllucination Detection Language Models Based on a Comprehensive Hallucination Taxonomy. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 6: Industry Track).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.acl-industry.11

@inproceedings{xu2026had,
  title = {{HAD: HAllucination Detection Language Models Based on a Comprehensive Hallucination Taxonomy}},
  author = {Fan Xu, Xinyu Hu, Zhenghan Yu, Li Lin, Xu Zhang, Yang Zhang, Wei Zhou, Jinjie Gu and Xiaojun Wan},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 6: Industry Track)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-industry.11},
  url = {https://doi.org/10.18653/v1/2026.acl-industry.11}
}

References

24 references, 13 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Forrest Bao, Miaoran Li, Rogger Luo, and Ofer Mendelevitch. 2024a. HHEM-2.1-Open. find
  2. Forrest Sheng Bao, Miaoran Li, Renyi Qu, Ge Luo, Erana Wan, Yujia Tang, Weisi Fan, Manveer Singh Tamber, Suleman Kazi, Vivek Sourabh, and 1 others. 2024b. Faithbench: A diverse hallucination benchmark for summarization by modern llms. arXiv preprint arXiv:2410.13210. find
  3. Shiqi Chen, Yiran Zhao, Jinghan Zhang, I-Chun Chern, Siyang Gao, Pengfei Liu, and Junxian He. 2023. Felm: Benchmarking factuality evaluation of large language models. In Thirty-seventh Conference on Neural Information Processing Systems Datasets and Benchmarks Track. find
  4. Xiang Chen, Duanzheng Song, Honghao Gui, Chengxi Wang, Ningyu Zhang, Jiang Yong, Fei Huang, Chengfei Lv, Dan Zhang, and Huajun Chen. 2024. Factchd: Benchmarking fact-conflicting hallucination detection. In Proceedings of the 33rd International Joint Conference on Artificial Intelligence. find
  5. Qinyuan Cheng, Tianxiang Sun, Wenwei Zhang, Siyin Wang, Xiangyang Liu, Mozhi Zhang, Junliang He, Mianqiu Huang, Zhangyue Yin, Kai Chen, and 1 others. 2023. Evaluating hallucinations in chinese large language models. arXiv preprint arXiv:2310.03368. — [arXiv:2310.03368](https://arxiv.org/abs/2310.03368) find
  6. Zina Chkirbene, Ridha Hamila, Ala Gouissem, and Unal Devrim. 2024. Large language models (llm) in industry: A survey of applications, challenges, and trends. In 2024 IEEE 21st International Conference on Smart Communities: Improving Quality of Life using AI, Robotics and IoT (HONET), pages 229–234.… find
  7. Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Mark Chen, Heewoo Jun, Lukasz Kaiser, Matthias Plappert, Jerry Tworek, Jacob Hilton, Reiichiro Nakano, Christopher Hesse, and John Schulman. 2021. Training verifiers to solve math word problems. arXiv preprint arXiv:2110.14168.
  8. David Dale, Elena Voita, Loïc Barrault, and Marta R Costa-Jussà. 2022. Detecting and mitigating hallucinations in machine translation: Model internal workings alone do well, sentence similarity even better. arXiv preprint arXiv:2212.08597.
  9. Nouha Dziri, Ehsan Kamalloo, Sivan Milton, Osmar Zaiane, Mo Yu, Edoardo M. Ponti, and Siva Reddy. 2022. FaithDial: A faithful benchmark for information-seeking dialogue. Transactions of the Association for Computational Linguistics, 10:1473– 1490.
  10. Angela Fan, Yacine Jernite, Ethan Perez, David Grangier, Jason Weston, and Michael Auli. 2019. ELI5: Long form question answering. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 3558–3567, Florence, Italy. Association for Computational Linguistics.
  11. Yuzhe Gu, Ziwei Ji, Wenwei Zhang, Chengqi Lyu, Dahua Lin, and Kai Chen. 2024. Anah-v2: Scaling analytical hallucination annotation of large language models. arXiv preprint arXiv:2407.04693. — [arXiv:2407.04693](https://arxiv.org/abs/2407.04693) find
  12. Lei Huang, Weijiang Yu, Weitao Ma, Weihong Zhong, Zhangyin Feng, Haotian Wang, Qianglong Chen, Weihua Peng, Xiaocheng Feng, Bing Qin, and 1 others. 2023. A survey on hallucination in large language models: Principles, taxonomy, challenges, and open questions. ACM Transactions on Information Systems.
  13. Gautier Izacard, Mathilde Caron, Lucas Hosseini, Sebastian Riedel, Piotr Bojanowski, Armand Joulin, and Edouard Grave. 2021. Unsupervised dense information retrieval with contrastive learning.
  14. Ziwei Ji, Nayeon Lee, Rita Frieske, Tiezheng Yu, Dan Su, Yan Xu, Etsuko Ishii, Ye Jin Bang, Andrea Madotto, and Pascale Fung. 2023. Survey of hallucination in natural language generation. ACM Computing Surveys, 55(12):1–38.
  15. Junyi Li, Xiaoxue Cheng, Xin Zhao, Jian-Yun Nie, and Ji-Rong Wen. 2023. HaluEval: A large-scale hallucination evaluation benchmark for large language models. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 6449–6464, Singapore. Association for Computa…
  16. Junyi Li, Tianyi Tang, Wayne Xin Zhao, Jian-Yun Nie, and Ji-Rong Wen. 2024. Pre-trained language models for text generation: A survey. ACM Computing Surveys, 56(9):1–39. find
  17. Potsawee Manakul, Adian Liusie, and Mark Gales. 2023. SelfCheckGPT: Zero-resource black-box hallucination detection for generative large language models. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 9004–9017, Singapore. Association for Computation… find
  18. Sewon Min, Kalpesh Krishna, Xinxi Lyu, Mike Lewis, Wen-tau Yih, Pang Koh, Mohit Iyyer, Luke Zettlemoyer, and Hannaneh Hajishirzi. 2023. FActScore: Fine-grained atomic evaluation of factual precision in long form text generation. In Proceedings of the 2023 Conference on Empirical Methods in Natural L…
  19. Abhika Mishra, Akari Asai, Vidhisha Balachandran, Yizhong Wang, Graham Neubig, Yulia Tsvetkov, and Hannaneh Hajishirzi. 2024. Fine-grained hallucination detection and editing for language models. arXiv preprint arXiv:2401.06855.
  20. Cheng Niu, Yuanhao Wu, Juno Zhu, Siliang Xu, Kashun Shum, Randy Zhong, Juntong Song, and Tong Zhang. 2023. Ragtruth: A hallucination corpus for developing trustworthy retrieval-augmented language models. arXiv preprint arXiv:2401.00396.
  21. Kellin Pelrine, Mohammad Taufeeque, Michał Zaj ac, Euan McLean, and Adam Gleave. 2023. Exploiting novel gpt-4 apis. arXiv preprint arXiv:2312.14302. — [arXiv:2312.14302](https://arxiv.org/abs/2312.14302) find
  22. Vikas Raunak, Arul Menezes, and Marcin Junczys-Dowmunt. 2021. The curious case of hallucinations in neural machine translation. arXiv preprint arXiv:2104.06683.
  23. Selvan Sunitha Ravi, Bartosz Mielczarek, Anand Kannappan, Douwe Kiela, and Rebecca Qian. 2024. Lynx: An open source hallucination evaluation model. arXiv preprint arXiv:2407.08488. — [arXiv:2407.08488](https://arxiv.org/abs/2407.08488) find
  24. Jeffrey Zhou, Tianjian Lu, Swaroop Mishra, Siddhartha Brahma, Sujoy Basu, Yi Luan, Denny Zhou, and Le Hou. 2023. Instruction-following evaluation for large language models. arXiv preprint arXiv:2311.07911.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/xu-2026-had-hallucination-detection-language-models/