Elamur Research⌕
☰
home / papers / 2026 / Streaming Hallucination Detection in Long Chain-of-Thought Reasoning

Streaming Hallucination Detection in Long Chain-of-Thought Reasoning

Paper · 2026 · Haolang Lu, Minghui Pan, Ripeng LI et al.

Summary

Errors in long chain-of-thought (CoT) reasoning tend to appear gradually and spread across steps, often hidden behind locally convincing text. Most detectors only check the final answer or individual steps. This paper treats hallucination as a latent state that changes as reasoning unfolds, not as a single faulty event. The authors use two signals. A step-level probe on the model's hidden states flags whether the current sentence adds unsupported content. A lightweight prefix-level predictor combines this local evidence with the hidden state to judge whether the reasoning so far has become hallucinated. They also show that standard additive pooling of token states harms step-level probing.

To train and test this, they built a dataset of over 10,000 annotated trajectories from LLaMA, Qwen and DeepSeek. Both levels reach above 87% accuracy, and the method detects hallucinations online with no extra inference cost. This matters because it gives real-time, interpretable warnings that reasoning is going off track. However, it requires white-box access to the model's internal states.

Key points

  • Dataset scale and sources: over 200k reasoning steps, including 40k+ hallucinated ones.
    • The queries are taken from existing benchmarks, namely those introduced by Suzgun et al. (2023) and Trivedi et al. (2022).
    • Average CoT length ranges from about 18 steps (Qwen) to about 32 steps (DeepSeek).
  • Annotation and validation:
    • Claude 4.5 assigned the step and prefix labels.
    • Logical checks required the final prefix state to match answer correctness and ruled out impossible state transitions.
    • Human review agreed with the automatic labels in over 96% of cases.
  • Probing design: each sentence counts as one step. The authors identify two flaws of additive token aggregation: cross-step saturation and within-step imbalance.
  • Results: above 87% accuracy for both step-level and prefix-level detection. 78% of CoT instances are correctly flagged during streaming.
  • Analysis: the authors propose eight logic-based dynamic metrics and report seven observations. One example is that prefix-level hallucination rises quickly but falls only after sustained corrective steps.

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

🎓Nanyang Technological UniversitySingapore · Singapore156 works · 25,644 citations🎓Renmin University of ChinaBeijing · China114 works · 16,118 citations🎓Beijing University of Posts and TelecommunicationsBeijing · China57 works · 32,843 citations🎓Southwest Jiaotong UniversityChengdu · China2 works · 402 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

🧭 Topics 1Research subjects and languages the work is about.Reasoning · 1,426 works

Cite

Cite as Haolang Lu, Minghui Pan, Ripeng LI, Guoshun Nan, Jialin Zhuang, Zijie Zhao, Zhongxiang Sun, Kun Wang and Yang Liu. 2026. Streaming Hallucination Detection in Long Chain-of-Thought Reasoning. 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.1064

@inproceedings{lu2026streaming,
  title = {{Streaming Hallucination Detection in Long Chain-of-Thought Reasoning}},
  author = {Haolang Lu, Minghui Pan, Ripeng LI, Guoshun Nan, Jialin Zhuang, Zijie Zhao, Zhongxiang Sun, Kun Wang and Yang Liu},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.1064},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.1064}
}

References

52 references, 22 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Zeyuan Allen-Zhu and Yuanzhi Li. 2024. Physics of Language Models: Part 3.1, Knowledge Storage and Extraction. In Proceedings of the 41st International Conference on Machine Learning.
  2. Amos Azaria and Tom Mitchell. 2023. The internal state of an llm knows when it’s lying. In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 967–976.
  3. Yuntai Bao, Xuhong Zhang, Tianyu Du, Xinkui Zhao, Zhengwen Feng, Hao Peng, and Jianwei Yin. 2025. Probing the geometry of truth: Consistency and generalization of truth directions in LLMs across logical transformations and question answering tasks. In Findings of the Association for Computational Li… find
  4. Yonatan Belinkov. 2022. Probing classifiers: Promises, shortcomings, and advances. Computational Linguistics, 48(1):207–219.
  5. Lennart Bürger, Fred A Hamprecht, and Boaz Nadler. 2024. Truth is universal: Robust detection of lies in llms. Advances in Neural Information Processing Systems, 37:138393–138431. find
  6. Zhiyu Chen, Wenhu Chen, Charese Smiley, Sameena Shah, Iana Borova, Dylan Langdon, Reema Moussa, Matt Beane, Ting-Hao Huang, Bryan R Routledge, and 1 others. 2021. Finqa: A dataset of numerical reasoning over financial data. In Proceedings of the 2021 Conference on Empirical Methods in Natural Langua…
  7. Jiahao Cheng, Tiancheng Su, Jia Yuan, Guoxiu He, Jiawei Liu, Xinqi Tao, Jingwen Xie, and Huaxia Li. 2025a. Chain-of-thought prompting obscures hallucination cues in large language models: An empirical evaluation. In Findings of the Association for Computational Linguistics: EMNLP 2025, pages 1272–13… find
  8. Xiaoxue Cheng, Junyi Li, Wayne Xin Zhao, and Ji-Rong Wen. 2025b. Think more, hallucinate less: Mitigating hallucinations via dual process of fast and slow thinking. In Findings of the Association for Computational Linguistics: ACL 2025, pages 7979– 7990. find
  9. Ziwei Deng, Mian Deng, Chenjing Liang, Zeming Gao, Chennan Ma, Chenxing Lin, Haipeng Zhang, Songzhu Mei, Siqi Shen, and Cheng Wang. 2025. Planu: Large language model reasoning through planning under uncertainty. In The Thirty-ninth Annual Conference on Neural Information Processing Systems. find
  10. Yihe Dong, Jean-Baptiste Cordonnier, and Andreas Loukas. 2021. Attention is not all you need: Pure attention loses rank doubly exponentially with depth. In International conference on machine learning, pages 2793–2803. PMLR. find
  11. Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Yang, Angela Fan, and 1 others. 2024. The llama 3 herd of models. arXiv e-prints, pages arXiv–2407.
  12. Ekaterina Fadeeva, Aleksandr Rubashevskii, Artem Shelmanov, Sergey Petrakov, Haonan Li, Hamdy Mubarak, Evgenii Tsymbalov, Gleb Kuzmin, Alexander Panchenko, Timothy Baldwin, Preslav Nakov, and Maxim Panov. 2024. Fact-checking the output of large language models via token-level uncertainty quantificat…
  13. Sebastian Farquhar, Jannik Kossen, Lorenz Kuhn, and Yarin Gal. 2024. Detecting hallucinations in large language models using semantic entropy. Nature, pages 625–630. find
  14. Zeyu Gan, Yun Liao, and Yong Liu. 2025. Rethinking external slow-thinking: From snowball errors to probability of correct reasoning. In Proceedings of the 42nd International Conference on Machine Learning (ICML). find
  15. Aman Goel, Daniel Schwartz, and Yanjun Qi. 2025. Zero-knowledge LLM hallucination detection and mitigation through fine-grained cross-model consistency. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing: Industry Track, pages 1982–1999. find
  16. Jiayi He, Hehai Lin, Qingyun Wang, Yi R. Fung, and Heng Ji. 2025. Self-correction is more than refinement: A learning framework for visual and language reasoning tasks. In Findings of the Association for Computational Linguistics: ACL 2025, pages 6405– 6421. Association for Computational Linguistics… find
  17. Bairu Hou, Yang Zhang, Jacob Andreas, and Shiyu Chang. 2025. A probabilistic framework for LLM hallucination detection via belief tree propagation. In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologi…
  18. Lei Huang, Weijiang Yu, Weitao Ma, Weihong Zhong, Zhangyin Feng, Haotian Wang, Qianglong Chen, Weihua Peng, Xiaocheng Feng, Bing Qin, and 1 others. 2025. A survey on hallucination in large language models: Principles, taxonomy, challenges, and open questions. ACM Transactions on Information Systems,…
  19. Soyeong Jeong, Jinheon Baek, Sukmin Cho, Sung Ju Hwang, and Jong C. Park. 2024. Adaptive-rag: Learning to adapt retrieval-augmented large language models through question complexity. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics:…
  20. Mingjian Jiang, Yangjun Ruan, Prasanna Sattigeri, Salim Roukos, and Tatsunori Hashimoto. 2024. Graph-based uncertainty metrics for long-form language model outputs. In Advances in Neural Information Processing Systems, volume 37, pages 11649– 11685. find
  21. Di Jin, Eileen Pan, Nassim Oufattole, Wei-Hung Weng, Hanyi Fang, and Peter Szolovits. 2021. What disease does this patient have? a large-scale open domain question answering dataset from medical exams. Applied Sciences, 11(14):6421.
  22. Nelson F Liu, Matt Gardner, Yonatan Belinkov, Matthew E Peters, and Noah A Smith. 2019. Linguistic knowledge and transferability of contextual representations. In Proceedings of NAACL-HLT, pages 1073–1094.
  23. Haolang Lu, Yilian Liu, Jingxin Xu, Guoshun Nan, Yuanlong Yu, Zhican Chen, and Kun Wang. 2025. Auditing meta-cognitive hallucinations in reasoning large language models. arXiv preprint arXiv:2505.13143. — [arXiv:2505.13143](https://arxiv.org/abs/2505.13143) find
  24. Zhenjiang Mao, Artem Bisliouk, Rohith Reddy Nama, and Ivan Ruchkin. 2025. Temporalizing confidence: Evaluation of chain-of-thought reasoning with signal temporal logic. arXiv preprint arXiv:2506.08243. find
  25. Samuel Marks and Max Tegmark. 2024. The geometry of truth: Emergent linear structure in large language model representations of true/false datasets. In First Conference on Language Modeling.
  26. Xin Qiu and Risto Miikkulainen. 2024. Semantic density: Uncertainty quantification for large language models through confidence measurement in semantic space. In Advances in Neural Information Processing Systems, pages 134507–134533. find
  27. Maohao Shen, Subhro Das, Kristjan Greenewald, Prasanna Sattigeri, Gregory W. Wornell, and Soumya Ghosh. 2024. Thermometer: Towards universal calibration for large language models. In Proceedings of the 41st International Conference on Machine Learning, pages 44687–44711. PMLR. find
  28. Gaurang Sriramanan, Siddhant Bharti, Vinu Sankar Sadasivan, Shoumik Saha, Priyatham Kattakinda, and Soheil Feizi. 2024. Llm-check: Investigating detection of hallucinations in large language models. Advances in Neural Information Processing Systems, 37:34188–34216. find
  29. Jingran Su, Jingfan Chen, Hongxin Li, Yuntao Chen, Li Qing, and Zhaoxiang Zhang. 2025. Activation steering decoding: Mitigating hallucination in large vision-language models through bidirectional hidden state intervention. In Proceedings of the 63rd Annual Meeting of the Association for Computationa…
  30. Weihang Su, Changyue Wang, Qingyao Ai, Yiran Hu, Zhijing Wu, Yujia Zhou, and Yiqun Liu. 2024. Unsupervised real-time hallucination detection based on the internal states of large language models. In Findings of the Association for Computational Linguistics ACL 2024, pages 14379–14391.
  31. Yuan Sui, Yufei He, Nian Liu, Xiaoxin He, Kun Wang, and Bryan Hooi. 2025. Fidelis: Faithful reasoning in large language models for knowledge graph question answering. In Findings of the Association for Computational Linguistics: ACL 2025, pages 8315–8330. find
  32. Zhongxiang Sun, Qipeng Wang, Haoyu Wang, Xiao Zhang, and Jun Xu. 2025. Detection and mitigation of hallucination in large reasoning models: A mechanistic perspective. arXiv preprint arXiv:2505.12886. — [arXiv:2505.12886](https://arxiv.org/abs/2505.12886) find
  33. Malavika Suresh, Rahaf Aljundi, Ikechukwu Nkisi-Orji, and Nirmalie Wiratunga. 2025. Cross-layer attention probing for fine-grained hallucination detection. arXiv preprint arXiv:2509.09700. — [arXiv:2509.09700](https://arxiv.org/abs/2509.09700) find
  34. Mirac Suzgun, Nathan Scales, Nathanael Schärli, Sebastian Gehrmann, Yi Tay, Hyung Won Chung, Aakanksha Chowdhery, Quoc Le, Ed Chi, Denny Zhou, and 1 others. 2023. Challenging big-bench tasks and whether chain-of-thought can solve them. In Findings of the Association for Computational Linguistics: AC…
  35. Ian Tenney, Patrick Xia, Berlin Chen, Alex Wang, Adam Poliak, R Thomas McCoy, Najoung Kim, Benjamin Van Durme, Samuel R Bowman, Dipanjan Das, and 1 others. 2019. What do you learn from context? probing for sentence structure in contextualized word representations. In International Conference on Lear…
  36. Harsh Trivedi, Niranjan Balasubramanian, Tushar Khot, and Ashish Sabharwal. 2022. MuSiQue: Multi-hop questions via single-hop question composition. Transactions of the Association for Computational Linguistics, pages 539–554. find
  37. Ning Wang, Yuan-Chen Jiang, Ming-Hui Zhai, and Wen-Song Liu. 2025. Large language model of electric power and application based on qwen. In 2025 IEEE 2nd International Conference on Energy and Electrical Engineering (EEE), pages 1–11. IEEE. find
  38. Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Fei Xia, Ed Chi, Quoc V Le, Denny Zhou, and 1 others. 2022. Chain-of-thought prompting elicits reasoning in large language models. Advances in neural information processing systems, 35:24824– 24837.
  39. Shusheng Xu, Wei Fu, Jiaxuan Gao, Wenjie Ye, Weilin Liu, Zhiyu Mei, Guangju Wang, Chao Yu, and Yi Wu. 2024. Is dpo superior to ppo for llm alignment? a comprehensive study. In Proceedings of the 41st International Conference on Machine Learning, pages 54983–54998. find
  40. Zhilin Yang, Peng Qi, Saizheng Zhang, Yoshua Bengio, William Cohen, Ruslan Salakhutdinov, and Christopher D Manning. 2018. Hotpotqa: A dataset for diverse, explainable multi-hop question answering. In Proceedings of the 2018 conference on empirical methods in natural language processing, pages 2369–…
  41. Zijun Yao, Yantao Liu, Yanxu Chen, Jianhui Chen, Junfeng Fang, Lei Hou, Juanzi Li, and Tat-Seng Chua. 2025. Are reasoning models more prone to hallucination? arXiv preprint arXiv:2505.23646. — [arXiv:2505.23646](https://arxiv.org/abs/2505.23646) find
  42. Tian Ye, Zicheng Xu, Yuanzhi Li, and Zeyuan Allen-Zhu. 2025a. Physics of Language Models: Part 2.1, Grade-School Math and the Hidden Reasoning Process. In Proceedings of the 13th International Conference on Learning Representations. find
  43. Tian Ye, Zicheng Xu, Yuanzhi Li, and Zeyuan Allen-Zhu. 2025b. Physics of Language Models: Part 2.2, How to Learn From Mistakes on Grade-School Math Problems. In Proceedings of the 13th International Conference on Learning Representations. find
  44. Sheldon Yu, Yuxin Xiong, Junda Wu, Xintong Li, Tong Yu, Xiang Chen, Ritwik Sinha, Jingbo Shang, and Julian McAuley. 2025a. Explainable chain-of-thought reasoning: An empirical analysis on state-aware reasoning dynamics. arXiv preprint arXiv:2509.00190. find
  45. Yiyao Yu, Yuxiang Zhang, Dongdong Zhang, Xiao Liang, Hengyuan Zhang, Xingxing Zhang, Mahmoud Khademi, Hany Hassan Awadalla, Junjie Wang, Yujiu Yang, and 1 others. 2025b. Chain-of-reasoning: Towards unified mathematical reasoning in large language models via a multi-paradigm perspective. In Proceedin…
  46. Anqi Zhang, Yulin Chen, Jane Pan, Chen Zhao, Aurojit Panda, Jinyang Li, and He He. 2025a. Reasoning models know when they’re right: Probing hidden states for self-verification. arXiv preprint arXiv:2504.05419. find
  47. Fujie Zhang, Peiqi Yu, Biao Yi, Baolei Zhang, Tong Li, and Zheli Liu. 2025b. Prompt-guided internal states for hallucination detection of large language models. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 21806–21818.
  48. Luan Zhang, Dandan Song, Zhijing Wu, Yuhang Tian, Changzhi Zhou, Jing Xu, Ziyi Yang, and Shuhao Zhang. 2025c. Detecting hallucination in large language models through deep internal representation analysis. In Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence,… find
  49. Yue Zhang, Yafu Li, Leyang Cui, Deng Cai, Lemao Liu, Tingchen Fu, Xinting Huang, Enbo Zhao, Yu Zhang, Yulong Chen, and 1 others. 2025d. Siren’s song in the ai ocean: A survey on hallucination in large language models. Computational Linguistics, pages 1–46.
  50. Zhenliang Zhang, Xinyu Hu, Huixuan Zhang, Junzhe Zhang, and Xiaojun Wan. 2025e. Icr probe: Tracking hidden state dynamics for reliable hallucination detection in llms. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 17986– 180…
  51. Yuqi Zhou, Sunhao Dai, Zhanshuo Cao, Xiao Zhang, and Jun Xu. 2025. Length-induced embedding collapse in plm-based models. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 28767–28791. find
  52. This appendix provides supplementary details to support the methodology and experimental results presented in the main text. 1. Appendix A elaborates on the rigorous data validation mechanism used to ensure dataset quality. 2. Appendix B provides a theoretical analysis of representation bias and the… find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/lu-2026-streaming-hallucination-detection-long-chain/