Elamur Research⌕
☰
home / papers / 2026 / Fine-Grained Detection of Context-Grounded Hallucinations Using LLMs

Fine-Grained Detection of Context-Grounded Hallucinations Using LLMs

Paper · 2026 · Yehonatan Peisakhovsky, Zorik Gekhman, Yosi Mass et al.

Summary

When models generate text from a given source, such as summaries or retrieval-augmented answers, they sometimes add claims that the source cannot support. Most existing checkers either give one consistent/inconsistent verdict for the whole output or rely on multi-stage pipelines. Those pipelines mark errors with entities, spans, QA pairs or atomic facts, which cannot express every kind of mistake. The authors test whether an LLM alone can locate such unsupported content end to end. They build FINAL, a benchmark derived from DeFacto summarization data and re-annotated by experts with LLM assistance. Each error is recorded as a short free-text explanation rather than a highlighted span. An LLM judge, validated against human ratings, matches predicted explanations to the gold ones. Four frontier models were tested under several prompting setups, and the best reached an F1 of only 0.67. The error analysis found two main failure modes. Models flag omitted details as contradictions, and they accept unverifiable statements that happen to match their own world knowledge. The resource offers a reproducible way to compare LLMs as fine-grained hallucination detectors.

Key points

  • FINAL (Factual Inconsistencies Localization) contains 1,400 annotated examples built on DeFacto (Liu et al., 2023), and its human annotation covered over 1,000 examples. A two-phase process combining human and LLM work raised the number of discovered inconsistencies by 31%.
  • Models evaluated: Llama-3-405B, GPT-4o, Gemini-Pro and Claude-Sonnet. The top F1 was 0.67. A Pipeline baseline adapted from Min et al. (2023) was also included.
  • Prompting with reasoning helps. Two-step setups, which first classify the output and then list errors, underperform end-to-end localization because the binary step is too conservative and recall drops.
  • Main failure sources:
    • flagging missing information as inconsistent, despite instructions to check only facts present in the output;
    • treating correct but unsupported facts as consistent because they align with the model's parametric knowledge.
  • Stated limitations: the benchmark covers only summarization with short documents, it is designed specifically for LLM outputs, and some errors may remain unannotated. Data and code are released on GitHub (yonip97/The_final_benchmark).

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

🎓Technion – Israel Institute of TechnologyHaifa · Israel49 works · 5,306 citations

 Fields, topics and concepts

📚 Shelves 6The 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 worksSocial Sciences · 1,772 worksSociology and Political Science · 577 works

Cite

Cite as Yehonatan Peisakhovsky, Zorik Gekhman, Yosi Mass, Liat Ein-Dor and Roi Reichart. 2026. Fine-Grained Detection of Context-Grounded Hallucinations Using LLMs. Findings of the Association for Computational Linguistics: ACL 2026.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.findings-acl.1907

@inproceedings{peisakhovsky2026fine,
  title = {{Fine-Grained Detection of Context-Grounded Hallucinations Using LLMs}},
  author = {Yehonatan Peisakhovsky, Zorik Gekhman, Yosi Mass, Liat Ein-Dor and Roi Reichart},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.1907},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.1907}
}

References

45 references, 26 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, et al. 2023. Gpt-4 technical report. arXiv preprint arXiv:2303.08774.
  2. Anthropic. 2024. Claude ai models. https://docs.anthropic.com/en/docs/welcome. ↗
  3. Ge Bai, Jie Liu, Xingyuan Bu, Yancheng He, Jiaheng Liu, Zhanhui Zhou, Zhuoran Lin, Wenbo Su, Tiezheng Ge, Bo Zheng, et al. 2024. Mt-bench-101: A fine-grained benchmark for evaluating large language models in multi-turn dialogues. In Proceedings of the 62nd Annual Meeting of the Association for Compu…
  4. Bernd Bohnet, Vinh Q Tran, Pat Verga, Roee Aharoni, Daniel Andor, Livio Baldini Soares, Massimiliano Ciaramita, Jacob Eisenstein, Kuzman Ganchev, Jonathan Herzig, et al. 2022. Attributed question answering: Evaluation and modeling for attributed large language models. arXiv preprint arXiv:2212.08037…
  5. Meng Cao, Yue Dong, and Jackie Chi Kit Cheung. 2022. Hallucinated but factual! inspecting the factuality of hallucinations in abstractive summarization. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 3340–3354.
  6. Shuyang Cao and Lu Wang. 2021. Cliff: Contrastive learning for improving faithfulness and factuality in abstractive summarization. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 6633–6649.
  7. Arie Cattan, Paul Roit, Shiyue Zhang, David Wan, Roee Aharoni, Idan Szpektor, Mohit Bansal, and Ido Dagan. 2024. Localizing factual inconsistencies in attributable text generation. arXiv preprint arXiv:2410.07473.
  8. Mert Cemri, Melissa Z Pan, Shuyi Yang, Lakshya A Agrawal, Bhavya Chopra, Rishabh Tiwari, Kurt Keutzer, Aditya Parameswaran, Dan Klein, Kannan Ramchandran, et al. 2025. Why do multi-agent llm systems fail? In The Thirty-ninth Annual Conference on Neural Information Processing Systems Datasets and Ben… find
  9. Sihao Chen, Senaka Buthpitiya, Alex Fabrikant, Dan Roth, and Tal Schuster. 2023. Propsegment: A large-scale corpus for proposition-level segmentation and entailment recognition. In Findings of the Association for Computational Linguistics: ACL 2023, pages 8874–8893.
  10. Elizabeth Clark, Shruti Rijhwani, Sebastian Gehrmann, Joshua Maynez, Roee Aharoni, Vitaly Nikolaev, Thibault Sellam, Aditya Siddhant, Dipanjan Das, and Ankur Parikh. 2023. Seahorse: A multilingual, multifaceted dataset for summarization evaluation. In Proceedings of the 2023 Conference on Empirical… find
  11. Alexander Richard Fabbri, Chien-Sheng Wu, Wenhao Liu, and Caiming Xiong. 2022. Qafacteval: Improved qa-based factual consistency evaluation for summarization. In Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technolo…
  12. Constanza Fierro, Ruchira Dhar, Filippos Stamatiou, Nicolas Garneau, and Anders Søgaard. 2024. Defining knowledge: Bridging epistemology and large language models. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 16096–16111, Miami, Florida, USA. Assoc… find
  13. Zorik Gekhman, Eyal Ben David, Hadas Orgad, Eran Ofek, Yonatan Belinkov, Idan Szpektor, Jonathan Herzig, and Roi Reichart. 2025. Inside-out: Hidden factual knowledge in llms. arXiv e-prints, pages arXiv–2503. find
  14. Zorik Gekhman, Jonathan Herzig, Roee Aharoni, Chen Elkind, and Idan Szpektor. 2023. Trueteacher: Learning factual consistency evaluation with large language models. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 2053–2070.
  15. Zorik Gekhman, Gal Yona, Roee Aharoni, Matan Eyal, Amir Feder, Roi Reichart, and Jonathan Herzig. 2024. Does fine-tuning LLMs on new knowledge encourage hallucinations? In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 7765–7784, Miami, Florida, USA. As… find
  16. Akshay Goel, Almog Gueta, Omry Gilon, Chang Liu, Sofia Erell, Lan Huong Nguyen, Xiaohong Hao, Bolous Jaber, Shashir Reddy, Rupesh Kartha, et al. 2023. Llms accelerate annotation for medical information extraction. In Machine Learning for Health (ML4H), pages 82–100. PMLR. find
  17. Google. 2024. Gemini pro. https://ai.google.dev. ↗
  18. Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, et al. 2024. The llama 3 herd of models. arXiv preprint arXiv:2407.21783.
  19. Horace He and Thinking Machines Lab. 2025. Defeating nondeterminism in llm inference. Thinking Machines Lab: Connectionism. Https://thinkingmachines.ai/blog/defeating-nondeterminism-in-llm-inference/. find
  20. Or Honovich, Roee Aharoni, Jonathan Herzig, Hagai Taitelbaum, Doron Kukliansy, Vered Cohen, Thomas Scialom, Idan Szpektor, Avinatan Hassidim, and Yossi Matias. 2022. True: Re-evaluating factual consistency evaluation. In Proceedings of the 2022 Conference of the North American Chapter of the Associa…
  21. Or Honovich, Leshem Choshen, Roee Aharoni, Ella Neeman, Idan Szpektor, and Omri Abend. 2021. Q2: Evaluating factual consistency in knowledge-grounded dialogues via question generation and question answering. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, p… find
  22. Philippe Laban, Tobias Schnabel, Paul Bennett, and Marti A Hearst. 2022. Summac: Re-visiting nli-based models for inconsistency detection in summarization. Transactions of the Association for Computational Linguistics, 10:163–177. find
  23. Yuho Lee, Taewon Yun, Jason Cai, Hang Su, and Hwanjun Song. 2024. Unisumeval: Towards unified, fine-grained, multi-dimensional summarization evaluation for llms. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 3941–3960. find
  24. Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, et al. 2020. Retrieval-augmented generation for knowledge-intensive nlp tasks. Advances in neural information processing systems, 33:9459–9474.
  25. Junyi Li, Jie Chen, Ruiyang Ren, Xiaoxue Cheng, Wayne Xin Zhao, Jian-Yun Nie, and Ji-Rong Wen. 2024. The dawn after the dark: An empirical study on factuality hallucination in large language models. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1:…
  26. Yixin Liu, Budhaditya Deb, Milagro Teruel, Aaron Halfaker, Dragomir Radev, and Ahmed Hassan. 2023. On improving summarization factual consistency from natural language feedback. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages… find
  27. Zheheng Luo, Qianqian Xie, and Sophia Ananiadou. 2023. Chatgpt as a factual inconsistency evaluator for text summarization. arXiv preprint arXiv:2303.15621.
  28. Joshua Maynez, Shashi Narayan, Bernd Bohnet, and Ryan McDonald. 2020. On faithfulness and factuality in abstractive summarization. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 1906–1919.
  29. Saurav Kadavath, Tom Conerly, Amanda Askell, Tom Henighan, Dawn Drain, Ethan Perez, Nicholas Schiefer, Zac Hatfield-Dodds, Nova DasSarma, Eli Tran-Johnson, et al. 2022. Language models (mostly) know what they know. CoRR. find
  30. Sewon Min, Kalpesh Krishna, Xinxi Lyu, Mike Lewis, Wen-tau Yih, Pang Wei Koh, Mohit Iyyer, Luke Zettlemoyer, and Hannaneh Hajishirzi. 2023. Factscore: Fine-grained atomic evaluation of factual precision in long form text generation. arXiv preprint arXiv:2305.14251.
  31. Ryo Kamoi, Tanya Goyal, and Greg Durrett. 2023a. Shortcomings of question answering based factuality frameworks for error localization. In Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics, pages 132–146. find
  32. Yifei Ming, Senthil Purushwalkam, Shrey Pandit, Zixuan Ke, Xuan-Phi Nguyen, Caiming Xiong, and Shafiq Joty. 2025. Faitheval: Can your language model stay faithful to context, even if" the moon is made of marshmallows". In The Thirteenth International Conference on Learning Representations. find
  33. Ryo Kamoi, Tanya Goyal, Juan Diego Rodriguez, and Greg Durrett. 2023b. Wice: Real-world entailment for claims in wikipedia. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 7561–7583.
  34. 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.
  35. Reiichiro Nakano, Jacob Hilton, Suchir Balaji, Jeff Wu, Long Ouyang, Christina Kim, Christopher Hesse, Shantanu Jain, Vineet Kosaraju, William Saunders, et al. 2021. Webgpt: Browser-assisted question-answering with human feedback. arXiv preprint arXiv:2112.09332.
  36. Jiayi Ye, Yanbo Wang, Yue Huang, Dongping Chen, Qihui Zhang, Nuno Moniz, Tian Gao, Werner Geyer, Chao Huang, Pin-Yu Chen, et al. 2025. Justice or prejudice? quantifying biases in llm-as-a-judge. In International Conference on Learning Representations.
  37. Cheng Niu, Yuanhao Wu, Juno Zhu, Siliang Xu, Kashun Shum, Randy Zhong, Juntong Song, and Tong Zhang. 2024. Ragtruth: A hallucination corpus for developing trustworthy retrieval-augmented language models. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volu…
  38. Jingqing Zhang, Yao Zhao, Mohammad Saleh, and Peter Liu. 2020. PEGASUS: Pre-training with extracted gap-sentences for abstractive summarization. In Proceedings of the 37th International Conference on Machine Learning, volume 119 of Proceedings of Machine Learning Research, pages 11328–11339. PMLR.
  39. OpenAI. 2022. Chatgpt. https://openai.com/blog/ chatgpt. Model: gpt-3.5-turbo. ↗
  40. Lianmin Zheng, Wei-Lin Chiang, Ying Sheng, Siyuan Zhuang, Zhanghao Wu, Yonghao Zhuang, Zi Lin, Zhuohan Li, Dacheng Li, Eric Xing, et al. 2023. Judging llm-as-a-judge with mt-bench and chatbot arena. Advances in neural information processing systems, 36:46595–46623.
  41. Liyan Tang, Tanya Goyal, Alex Fabbri, Philippe Laban, Jiacheng Xu, Semih Yavuz, Wojciech Kryściński, Justin Rousseau, and Greg Durrett. 2023. Understanding factual errors in summarization: Errors, summarizers, datasets, error detectors. In Proceedings of the 61st Annual Meeting of the Association…
  42. Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, et al. 2023. Llama 2: Open foundation and fine-tuned chat models. arXiv preprint arXiv:2307.09288.
  43. Alex Wang, Kyunghyun Cho, and Mike Lewis. 2020. Asking and answering questions to evaluate the factual consistency of summaries. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 5008–5020. find
  44. Peiyi Wang, Lei Li, Liang Chen, Zefan Cai, Dawei Zhu, Binghuai Lin, Yunbo Cao, Lingpeng Kong, Qi Liu, Tianyu Liu, et al. 2024. Large language models are not fair evaluators. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 9440… find
  45. Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Fei Xia, Ed Chi, Quoc V Le, Denny Zhou, et al. 2022. Chain-of-thought prompting elicits reasoning in large language models. Advances in neural information processing systems, 35:24824–24837.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/peisakhovsky-2026-fine-grained-detection-context-grounded/