A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questions
Open accessThe DOI matches its Crossref record; 496 of the 496 references it cites were found in a registry or an index; 110 of the 234 sources its claims cite were checked for support.
Abstract
The emergence of large language models (LLMs) has marked a significant breakthrough in natural language processing (NLP), fueling a paradigm shift in information acquisition. Nevertheless, LLMs are prone to hallucination, generating plausible yet nonfactual content. This phenomenon raises significant concerns over the reliability of LLMs in real-world information retrieval (IR) systems and has attracted intensive research to detect and mitigate such hallucinations. Given the open-ended general-purpose attributes inherent to LLMs, LLM hallucinations present distinct challenges that diverge from prior task-specific models. This divergence highlights the urgency for a nuanced understanding and comprehensive overview of recent advances in LLM hallucinations. In this survey, we begin with an innovative taxonomy of hallucination in the era of LLM and then delve into the factors contributing to hallucinations. Subsequently, we present a thorough overview of hallucination detection methods and benchmarks. Our discussion then transfers to representative methodologies for mitigating LLM hallucinations. Additionally, we delve into the current limitations faced by retrieval-augmented LLMs in combating hallucinations, offering insights for developing more robust IR systems. Finally, we highlight the promising research directions on LLM hallucinations, including hallucination in large vision-language models and understanding of knowledge boundaries in LLM hallucinations.
Overview
CoChat Summary
TL;DR
This survey proposes a taxonomy of LLM hallucination into factuality and faithfulness, then reviews causes, detection methods, benchmarks, mitigation strategies, and RAG-specific challenges. It also outlines open directions including hallucinations in large vision-language models and understanding knowledge boundaries in LLMs.
Key findings
- Redefines LLM hallucination as factuality hallucination (factual contradiction: entity-error, relation-error; factual fabrication: unverifiability, overclaim) and faithfulness hallucination (instruction inconsistency, context inconsistency, logical inconsistency).
- Identifies causes across data (misinformation/biases, knowledge boundary, inferior alignment data), training (pre-training, SFT, RLHF), and inference (imperfect decoding, over-confidence, softmax bottleneck, reasoning failure).
- Reviews detection and benchmarks: factuality detection via fact-checking and uncertainty estimation; faithfulness detection via fact-based, classifier-based, QA-based, uncertainty-based, and LLM-based metrics. Reported benchmark sizes include TruthfulQA (817 questions), HaluEval (30,000 task-specific and 5,000 general samples), FreshQA (600 hand-crafted questions), and HaluEval 2.0 (8,770 questions across five domains).
- Mitigation is organized by cause: data filtering, model editing, RAG; pre-training/alignment fixes; and factuality/faithfulness enhanced decoding. RAG still faces retrieval failure and generation bottleneck.
Methods
- Design: a survey/overview that proposes a taxonomy and organizes literature on LLM hallucination principles, causes, detection, benchmarks, mitigation, RAG challenges, and future directions.
- Data or sample: no primary dataset or new experiments; synthesizes existing research and benchmarks such as TruthfulQA, HaluEval, FreshQA, and HaluEval 2.0.
- Main analysis: conceptual categorization and qualitative synthesis; no quantitative meta-analysis is reported.
Limitations
- The paper does not explicitly state limitations of the survey itself.
- It states limitations of reviewed methods: RAG can be impacted by irrelevant retrievals and shallow retriever-generator interactions; fact-checking solely via LLM parametric knowledge may yield inaccurate assessments; classifier-based faithfulness metrics are constrained by scarcity of annotated data.
- Text makes evident: as a survey, it presents no original empirical validation or new benchmark results.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
76 Valid · 5 Not valid · 48 Not checked · 108 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Recent studies have highlighted a strong correlation between the model's accuracy on general domain questions and the volume of relevant documents [145] or entity popularity [204] within the pre-training corpora.”
What the sources say
- Supportedcites [145] [145] Nikhil Kandpal, Haikang Deng, Adam Roberts, Eric Wallace, and Colin Raffel. 2023. Large Language Models Struggle to Learn Long-Tail Knowledge. In International Conference on Machine Learning, ICML 2023, 23-29 July 2023, Honolulu, Hawaii, USA (Proceedings of Machine Learning Research, Vol. 202) , Andreas Krause, Emma Brunskill, Kyunghyun Cho, Barbara Engelhardt, Sivan Sabato, and Jonathan Scarlett (Eds.). PMLR, 15696-15707. https: //proceedings.mlr.press/v202/kandpal23a.html - source
The citation to this paper is attached to the volume of relevant documents, not the separate entity-popularity claim. Section 3.1 (p. 5) reports a strong correlation between question-answering accuracy and relevant pre-training document count across the tested models. (LLM, conf=0.94)
- Partly supportedcites [204] [204] Alex Mallen, Akari Asai, Victor Zhong, Rajarshi Das, Daniel Khashabi, and Hannaneh Hajishirzi. 2023. When Not to Trust Language Models: Investigating Effectiveness of Parametric and Non-Parametric Memories. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), ACL 2023, Toronto, Canada, July 9-14, 2023 , Anna Rogers, Jordan L. Boyd-Graber, and Naoaki Okazaki (Eds.). Association for Computational Linguistics, 9802-9822. https://doi.org/10.18653/v1/2023.acl-long.546 - sourceExperimental — accuracy not yet measured
The “Subject entity popularity predicts memorization” section reports a positive correlation between entity popularity and question-answering accuracy for almost all relationship types. However, the paper measures popularity using Wikipedia page views as a proxy, not entity frequency within pre-training corpora. Its Limitations section also says the experiments rely on synthetic datasets and that applicability to naturally occurring questions is not firmly established. (LLM, conf=0.87)
boostedstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Such biases can be inadvertently acquired from internet-based texts, which are rife with diverse and biased viewpoints, and subsequently be propagated into the generated content [155].”
What the sources say
- Partly supportedcites [155] [155] Faisal Ladhak, Esin Durmus, Mirac Suzgun, Tianyi Zhang, Dan Jurafsky, Kathleen McKeown, and Tatsunori Hashimoto. 2023. When Do Pre-Training Biases Propagate to Downstream Tasks? A Case Study in Text Summarization. In Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics . Association for Computational Linguistics, Dubrovnik, Croatia, 3206-3219. https://aclanthology.org/2023.eacl-main.234 - sourceExperimental — accuracy not yet measured
The Abstract and §6 Discussion report that pre-training biases can propagate into generated summaries as hallucinated facts. But §6 leaves the source of those biases for future work and treats stereotyping in pre-training data as one possibility; it does not establish that they were acquired from internet-based texts. (LLM, conf=0.86)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Neural networks possess an intrinsic tendency to memorize training data [35]”
What the sources say
- Partly supportedcites [35] [35] Nicholas Carlini, Florian Tramer, Eric Wallace, Matthew Jagielski, Ariel Herbert-Voss, Katherine Lee, Adam Roberts, Tom Brown, Dawn Song, Ulfar Erlingsson, et al. 2021. Extracting training data from large language models. In 30th USENIX Security Symposium (USENIX Security 21) . 2633-2650. - sourceExperimental — accuracy not yet measured
Section 3.1 (p. 3) says memorization is in many ways an essential component of language models, and the Introduction (p. 1) demonstrates that large language models memorize individual training examples. The claim extends that finding to neural networks generally; the cited passages do not establish an intrinsic tendency across all neural networks. (LLM, conf=0.83)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“However, certain biases are intrinsically tied to hallucinations, especially those related to gender [242] and nationality [155, 226].”
What the sources say
- Not matchedcites [242]
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Supportedcites [155] [155] Faisal Ladhak, Esin Durmus, Mirac Suzgun, Tianyi Zhang, Dan Jurafsky, Kathleen McKeown, and Tatsunori Hashimoto. 2023. When Do Pre-Training Biases Propagate to Downstream Tasks? A Case Study in Text Summarization. In Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics . Association for Computational Linguistics, Dubrovnik, Croatia, 3206-3219. https://aclanthology.org/2023.eacl-main.234 - source
The Introduction and Section 3.3 report a strong association between pre-trained models’ name–nationality bias and nationality hallucinations in downstream text summarization. The citation to this paper is attached to nationality; the separate gender citation is not assessed here. (LLM, conf=0.89)
- Not matchedcites [226]
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Not matched
Every source the paper cites for this claim was checked and supports it.
In the paper
“this memorization tendency grows with model size [34, 54]”
What the sources say
- Supportedcites [34] [34] Nicholas Carlini, Daphne Ippolito, Matthew Jagielski, Katherine Lee, Florian Tramer, and Chiyuan Zhang. 2022. Quantifying memorization across neural language models. ArXiv preprint abs/2202.07646 (2022). https://arxiv.org/ abs/2202.07646 - source
Section 4.1 (p. 4) reports that larger models memorize significantly more than smaller models. The Introduction (p. 2) specifies that this finding is within model families, and the Conclusion (p. 9) describes memorization as scaling with model size. (LLM, conf=0.94)
- Supportedcites [54] [54] Aakanksha Chowdhery, Sharan Narang, Jacob Devlin, Maarten Bosma, Gaurav Mishra, Adam Roberts, Paul Barham, Hyung Won Chung, Charles Sutton, Sebastian Gehrmann, Parker Schuh, Kensen Shi, Sasha Tsvyashchenko, Joshua Maynez, Abhishek Rao, Parker Barnes, Yi Tay, Noam Shazeer, Vinodkumar Prabhakaran, Emily Reif, Nan Du, Ben Hutchinson, Reiner Pope, James Bradbury, Jacob Austin, Michael Isard, Guy Gur-Ari, Pengcheng Yin, Toju Duke, Anselm Levskaya, Sanjay Ghemawat, Sunipa Dev, Henryk Michalewski, Xavier Garcia, Vedant Misra, Kevin Robinson, Liam Fedus, Denny Zhou, Daphne Ippolito, David Luan, H source
Section 7, “Memorization” (p. 35), explicitly concludes that larger models have a higher memorization rate than smaller models. It also reports that the exact-continuation match rate rises from 1.6% for the 8B model to 2.4% for the 540B model. (LLM, conf=0.94)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
102 more claims · 41 with their cited sources checked
Create a free account to see all 237 claims and to verify the 13 references still open.
- Probably not a substantial claim. 97
- It may state two separate claims as one. 55
- The quote only partly states this claim. 1
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“However, LLMs exhibit hallucination and fabrication errors that are unacceptable for quantitative modeling [19, 20, 21].”
From Structured Schemas for Provenance‐Rich, LLM ‐Assisted QSP Model Calibration (2026) · cites this paper as [20]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“In particular, the generation of non-existent academic references or legal precedents has been widely recognized as a critical issue [4].”
From Hallucinations in Bibliographic Recommendation: Citation Frequency as a Proxy for Training Data Redundancy (2025) · cites this paper as [4]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Retrieval augmentation, tool use, citation systems, constrained decoding, domain-specific training, and multimodal grounding can all supply partial external anchoring, and they alter both the frequency and the detectability of unsupported claims [15,29].”
From Hallucination as epistemic pollution: epistemic injustice and credibility in futures-oriented science education (2026) · cites this paper as [15]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“These measures do reduce the frequency of unsupported claims and improve the odds that a learner can detect one, and the argument here does not depend on denying it [15,29].”
From Hallucination as epistemic pollution: epistemic injustice and credibility in futures-oriented science education (2026) · cites this paper as [15]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Retrieval-augmented generation evaluation checks whether model claims are factually grounded in retrieved documents.It does not check whether retrieved evidence is attributed to the correct entity.”
From Deceptive Grounding: Entity Attribution Failure in Clinical Retrieval-Augmented Generation (2026) · cites this paper as Huang et al., 2025
Experimental — accuracy not yet measured
21 more claims about this paper
Create a free account to see every claim other papers make about this one.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 2.3 Hallucinations in Large Language Models · p. 7
“the final answer of x=4 is inconsistent with the reasoning chain, leading to an incorrect result.”
§ 2.3 Hallucinations in Large Language Models · p. 6
“Step 2: Divide both sides by 2 to get: x = 3. So the result of the equation is x=3.”
Citations
Cites
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019) · Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) · doi:10.18653/v1/n19-1423 · Cited by 54,325
Language Models are Few-Shot Learners
Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al. (2020) · arXiv · doi:10.48550/arxiv.2005.14165 · Cited by 23,372
BLEU: a method for automatic evaluation of machine translation
Papineni, K., Roukos, S., Ward, T., & Zhu, W.-J. (2001) · Proceedings of the 40th Annual Meeting on Association for Computational Linguistics - ACL '02 · doi:10.3115/1073083.1073135 · Cited by 21,344
RoBERTa: A Robustly Optimized BERT Pretraining Approach
Liu, Y., Ott, M., Goyal, N., Du, J., Joshi, M., Chen, D., Levy, O., Lewis, M., Zettlemoyer, L., & Stoyanov, V. (2019) · arXiv · doi:10.48550/arxiv.1907.11692 · Cited by 16,515
Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
Raffel, C., Shazeer, N., Roberts, A., Lee, K., Narang, S., Matena, M., Zhou, Y., Li, W., & Liu, P. J. (2019) · arXiv · doi:10.48550/arxiv.1910.10683 · Cited by 10,929
Proximal Policy Optimization Algorithms
Schulman, J., Wolski, F., Dhariwal, P., Radford, A., & Klimov, O. (2017) · arXiv · doi:10.48550/arxiv.1707.06347 · Cited by 10,625 · cited 1×
“To align with human preferences, RLHF optimizes the LLM to generate outputs that maximize the reward provided by the trained preference model, typically employing a reinforcement learning algorithm, such as Proximal Policy Optimization (PPO)(Schulman et al., 2017).”
Scaling Laws for Neural Language Models
Kaplan, J., McCandlish, S., Henighan, T., Brown, T. B., Chess, B., Child, R., Gray, S., Radford, A., Wu, J., & Amodei, D. (2020) · arXiv · doi:10.48550/arxiv.2001.08361 · Cited by 9,550 · cited 1×
“While scaling up pre-training data substantially enhances the competencies of LLMs (Kaplan et al., 2020; Hoffmann et al., 2022), challenges arise in maintaining consistent data quality, which can potentially introduce misinformation and biases (Bender et al., 2021; Weidinger et al., 2021).”
Lewis, M., Liu, Y., Goyal, N., Ghazvininejad, M., Mohamed, A., Levy, O., Stoyanov, V., & Zettlemoyer, L. (2020) · Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/2020.acl-main.703 · Cited by 8,821 · cited 10×
“By retrieving evidence from external knowledge bases, RAG enables LLMs to be equipped with up-to-date knowledge and generate responses conditioning on relevant evidence.”
InfluentialLewis, M., Liu, Y., Goyal, N., Ghazvininejad, M., Mohamed, A., Levy, O., Stoyanov, V., & Zettlemoyer, L. (2019) · arXiv · doi:10.48550/arxiv.1910.13461 · Cited by 8,816
2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)
(2017) · doi:10.1109/cvpr35066.2017 · Cited by 7,141
Research works printed but not linked: 33
- ArXiv. 2023. arxiv dataset. Retrieved from https://www.kaggle.com/datasets/Cornell-University/arxiv/versions/134
- Samy Bengio, Oriol Vinyals, Navdeep Jaitly, and Noam Shazeer. 2015. Scheduled sampling for sequence prediction with recurrent neural networks. In Proceedings of the 28th International Conference on Neural Information Processing Systems. Corinna Cortes, Neil D. Lawrence, Daniel D. Lee, Masashi Sugiyama, and Roman Garnett (Eds.), 1171–1179. Retrieved from https://proceedings.neurips.cc/paper/2015/hash/e995f98d56967d946471af29d7bf99f1-Abstract.html
- Sebastian Borgeaud, Arthur Mensch, Jordan Hoffmann, Trevor Cai, Eliza Rutherford, Katie Millican, George van den Driessche, Jean-Baptiste Lespiau, Bogdan Damoc, Aidan Clark, et al. 2022. Improving language models by retrieving from trillions of tokens. In Proceedings of theInternational Conference on Machine Learning (ICML ’22). Kamalika Chaudhuri, Stefanie Jegelka, Le Song, Csaba Szepesvári, Gang Niu, and Sivan Sabato (Eds.), Proceedings of Machine Learning Research, Vol. 162, PMLR, 2206–2240. Retrieved from https://proceedings.mlr.press/v162/borgeaud22a.html
- 10.5555/3495724.3495883 · doi:10.5555/3495724.3495883
- 10.5555/3600270.3602050 · doi:10.5555/3600270.3602050
- Paul F. Christiano, Jan Leike, Tom B. Brown, Miljan Martic, Shane Legg, and Dario Amodei. 2017. Deep reinforcement learning from human preferences. In Proceedings of the 31st International Conference on Neural Information Processing Systems. Isabelle Guyon, Ulrike von Luxburg, Samy Bengio, Hanna M. Wallach, Rob Fergus, S. V. N. Vishwanathan, and Roman Garnett (Eds.), 4299–4307. Retrieved from https://proceedings.neurips.cc/paper/2017/hash/d5e2c0adad503c91f91df240d0cd4e49-Abstract.html
- Sumanth Dathathri, Andrea Madotto, Janice Lan, Jane Hung, Eric Frank, Piero Molino, Jason Yosinski, and Rosanne Liu. 2020. Plug and play language models: A simple approach to controlled text generation. In Proceedings of the 8th International Conference on Learning Representations (ICLR ’20). OpenReview.net. Retrieved from https://openreview.net/forum?id=H1edEyBKDS
- Bikash Gyawali, Lucas Anastasiou, and Petr Knoth. 2020. Deduplication of scholarly documents using locality sensitive hashing and word embeddings. In Proceedings of the 12th Language Resources and Evaluation Conference. European Language Resources Association, 901–910. Retrieved from https://aclanthology.org/2020.lrec-1.113
- Ari Holtzman, Jan Buys, Li Du, Maxwell Forbes, and Yejin Choi. 2020. The curious case of neural text degeneration. In Proceedings of the 8th International Conference on Learning Representations (ICLR ’20). OpenReview.net. Retrieved from https://openreview.net/forum?id=rygGQyrFvH
- Zeyu Huang, Yikang Shen, Xiaofeng Zhang, Jie Zhou, Wenge Rong, and Zhang Xiong. 2023d. Transformer-Patcher: One mistake worth one neuron. In Proceedings of the 11th International Conference on Learning Representations (ICLR ’23). OpenReview.net. Retrieved from https://openreview.net/pdf?id=4oYUGeGBPm
- Gautier Izacard Mathilde Caron Lucas Hosseini Sebastian Riedel Piotr Bojanowski Armand Joulin and Edouard Grave. 2022. Unsupervised dense information retrieval with contrastive learning. Transactions on Machine Learning Research 2022 (2022). Retrieved from https://openreview.net/forum?id=jKN1pXi7b0
- Nikhil Kandpal, Haikang Deng, Adam Roberts, Eric Wallace, and Colin Raffel. 2023. Large language models struggle to learn long-tail knowledge. In Proceedings of the International Conference on Machine Learning (ICML ’23). Andreas Krause, Emma Brunskill, Kyunghyun Cho, Barbara Engelhardt, Sivan Sabato, and Jonathan Scarlett (Eds.), Proceedings of Machine Learning Research, Vol. 202, PMLR, 15696–15707. Retrieved from https://proceedings.mlr.press/v202/kandpal23a.html
- Urvashi Khandelwal, Omer Levy, Dan Jurafsky, Luke Zettlemoyer, and Mike Lewis. 2020. Generalization through memorization: Nearest neighbor language models. In Proceedings of the 8th International Conference on Learning Representations (ICLR ’20). OpenReview.net. Retrieved from https://openreview.net/forum?id=HklBjCEKvH
- 10.5555/3600270.3601883 · doi:10.5555/3600270.3601883
- Balaji Lakshminarayanan, Alexander Pritzel, and Charles Blundell. 2017. Simple and scalable predictive uncertainty estimation using deep ensembles. In Proceedings of the 31st International Conference on Neural Information Processing Systems. Isabelle Guyon, Ulrike von Luxburg, Samy Bengio, Hanna M. Wallach, Rob Fergus, S. V. N. Vishwanathan, and Roman Garnett (Eds.), 6402–6413. Retrieved from https://proceedings.neurips.cc/paper/2017/hash/9ef2ed4b7fd2c810847ffa5fa85bce38-Abstract.html
- 10.5555/3600270.3602776 · doi:10.5555/3600270.3602776
- 10.5555/3495724.3496517 · doi:10.5555/3495724.3496517
- Jinming Li, Wentao Zhang, Tian Wang, Guanglei Xiong, Alan Lu, and Gerard Medioni. 2023. GPT4Rec: A generative framework for personalized recommendation and user interests interpretation. In Proceedings of the SIGIR Workshop on eCommerce Co-located with the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR 23). Surya Kallumadi, Yubin Kim, Tracy Holloway King, Shervin Malmasi, Maarten de Rijke, and Jacopo Tagliabue (Eds.), CEUR Workshop Proceedings, Vol. 3589, CEUR-WS.org. Retrieved from https://ceur-ws.org/Vol-3589/paper_2.pdf
- Chin-Yew Lin. 2004. ROUGE: A package for automatic evaluation of summaries. In Text Summarization Branches Out. Association for Computational Linguistics, 74–81. Retrieved from https://aclanthology.org/W04-1013
- 10.5555/3600270.3601532 · doi:10.5555/3600270.3601532
- Kevin Meng, Arnab Sen Sharma, Alex J. Andonian, Yonatan Belinkov, and David Bau. 2023. Mass-editing memory in a transformer. In Proceedings of the 11th International Conference on Learning Representations (ICLR ’23). OpenReview.net. Retrieved from https://openreview.net/pdf?id=MkbcAHIYgyS
- Eric Mitchell, Charles Lin, Antoine Bosselut, Chelsea Finn, and Christopher D. Manning. 2022. Fast model editing at scale. In Proceedings of the 10th International Conference on Learning Representations (ICLR ’22). OpenReview.net. Retrieved from https://openreview.net/forum?id=0DcZxeWfOPt
- Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll L. Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. 2022. Training language models to follow instructions with human feedback. In NeurIPS. Retrieved from http://papers.nips.cc/paper_files/paper/2022/hash/b1efde53be364a73914f58805a001731-Abstract-Conference.html
- Alec Radford, Jeffrey Wu, Rewon Child, David Luan, Dario Amodei, and Ilya Sutskever. 2019. Language models are unsupervised multitask learners. OpenAI blog 1, 8 (2019), 9.
- Anton Sinitsin, Vsevolod Plokhotnyuk, Dmitriy Pyrkin, Sergei Popov, and Artem Babenko. 2020. Editable neural networks. In Proceedings of the8th International Conference on Learning Representations (ICLR ’20). OpenReview.net. Retrieved from https://openreview.net/forum?id=HJedXaEtvS
- 10.5555/3495724.3495977 · doi:10.5555/3495724.3495977
- Chenmien Tan, Ge Zhang, and Jie Fu. 2024. Massive editing for large language model via meta learning. In Proceedings of the 12th International Conference on Learning Representations. Retrieved from https://openreview.net/forum?id=L6L1CJQ2PE
- Zirui Wang Jiahui Yu Adams Wei Yu Zihang Dai Yulia Tsvetkov and Yuan Cao. 2022. SimVLM: Simple visual language model pretraining with weak supervision. In Proceedings of the 10th International Conference on Learning Representations (ICLR ’22). OpenReview.net. Retrieved from https://openreview.net/forum?id=GUrhfTuf_3
- 10.5555/3600270.3602070 · doi:10.5555/3600270.3602070
- Zhilin Yang, Zihang Dai, Ruslan Salakhutdinov, and William W. Cohen. 2018. Breaking the softmax bottleneck: A high-rank RNN language model. In Proceedings of the6th International Conference on Learning Representations (ICLR ’18). OpenReview.net. Retrieved from https://openreview.net/forum?id=HkwZSG-CZ
- Zhilin Yang, Thang Luong, Ruslan Salakhutdinov, and Quoc V. Le. 2019. Mixtape: Breaking the softmax bottleneck efficiently. In Proceedings of the 33rd International Conference on Neural Information Processing Systems (NeurIPS ’19). Hanna M. Wallach, Hugo Larochelle, Alina Beygelzimer, Florence d’Alché-Buc, Emily B. Fox, and Roman Garnett (Eds.), 15922–15930. Retrieved from https://proceedings.neurips.cc/paper/2019/hash/512fc3c5227f637e41437c999a2d3169-Abstract.html
- Weizhe Yuan, Graham Neubig, and Pengfei Liu. 2021. BARTScore: Evaluating generated text as text generation. In Proceedings of the 35th International Conference on Neural Information Processing Systems (NeurIPS ’21). Marc’Aurelio Ranzato, Alina Beygelzimer, Yann N. Dauphin, Percy Liang, and Jennifer Wortman Vaughan (Eds.), 27263–27277. Retrieved from https://proceedings.neurips.cc/paper/2021/hash/e4d2b6e6fdeca3e60e0f1a62fee3d9dd-Abstract.html
- Hugh Zhang, Daniel Duckworth, Daphne Ippolito, and Arvind Neelakantan. 2021. Trading off diversity and quality in natural language generation. In Proceedings of the Workshop on Human Evaluation of NLP Systems (HumEval). Association for Computational Linguistics, 25–33. Retrieved from https://aclanthology.org/2021.humeval-1.3
Cited by
Items 201–300 of 2,243
Beyond Functional Correctness: Exploring Hallucinations in LLM-Generated Code
Liu, F., Liu, Y., Shi, L., Yang, Z., Zhang, L., Lian, X., Li, Z., & Ma, Y. (2026) · IEEE Transactions on Software Engineering · doi:10.1109/tse.2026.3657432 · Cited by 408
“Hallucinations refers to the phenomenon that LLMs might produce nonsensical or unfaithful outputs to the provided source content [24], [28], which can poses a potential risk in deploying LLMs across various applications.”
InfluentialAssessing the risk of takeover catastrophe from large language models
Baum, S. D. (2024) · Risk Analysis · doi:10.1111/risa.14353 · Cited by 11
“Much discussed are LLM “hallucinations,” in which the LLM fabricates false information (Huang et al., 2023).”
Swamped with Too Many Articles? GraphRAG Makes Getting Started Easy
Ngangmeni, J., & Rawat, D. B. (2025) · AI · doi:10.3390/ai6030047 · Cited by 13
“LLMs sometimes generate false information [7], but since it is becoming increasingly challenging to manually check their output as data pool sizes increase, using LLMs to judge other LLM output is now an open area of research [8].”
Lafourcade, C., Kérourédan, O., Ballester, B., & Richert, R. (2025) · Journal of Dentistry · doi:10.1016/j.jdent.2025.105764 · Cited by 20
Decoding the Molecular Language of Proteins with EvollaPreprint
Zhou, X., Han, C., Zhang, Y., Du, H., Tian, J., Su, J., Liu, R., Zhuang, K., Jiang, S., Gitter, A., et al. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.01.05.630192 · Cited by 39
AI assistance in peer feedback provision: Pedagogically sound, but minimally adopted
Pozdniakov, S., Brazil, J., Banihashem, S. K., Noroozi, O., Gašević, D., Sadiq, S., & Khosravi, H. (2026) · Computers & Education · doi:10.1016/j.compedu.2026.105591 · Cited by 11
The evolution of language models: From N-Grams to LLMs, and beyond
Raeini, M. (2025) · Natural Language Processing Journal · doi:10.1016/j.nlp.2025.100168 · Cited by 40
“Who will be left behind?”: A Swedish case of learning AI in vocational education
Metreveli, A., Chen, X., Hedman, A., & Sergeeva, A. (2025) · International Journal of Educational Research · doi:10.1016/j.ijer.2025.102697 · Cited by 24
IBD: Alleviating Hallucinations in Large Vision-Language Models via Image-Biased Decoding
Zhu, L., Ji, D., Chen, T., Xu, P., Ye, J., & Liu, J. (2025) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw67362.2025.00150 · Cited by 151
Pharmacometrics in the Age of Large Language Models: A Vision of the Future
Tosca, E. M., Aiello, L., Carlo, A. D., & Magni, P. (2025) · Pharmaceutics · doi:10.3390/pharmaceutics17101274 · Cited by 13
“However, directly using LLMs as a search engine for information retrieval carries the risk of hallucinations [64], i.e., the generation of plausible but incorrect statements fictitious citations [65].”
LLMs for Accessibility in Mobile Apps: Detection and Repair
Aljedaani, W., Aljohani, A., Eler, M. M., Habib, A., & Do, H. (2025) · International Web for All Conference (W4A) · doi:10.1145/3744257.3744270 · Cited by 12
KNighter: Transforming Static Analysis with LLM-Synthesized Checkers
Yang, C., Zhao, Z., Xie, Z., Li, H., & Zhang, L. (2025) · ACM SIGOPS Symposium on Operating Systems Principles (SOSP) · doi:10.1145/3731569.3764827 · Cited by 68
Shadiev, R., Nguyen, T. M., & Hwang, W. (2026) · Interactive Learning Environments · doi:10.1080/10494820.2025.2612252 · Cited by 17
“As McIntosh et al. (2023) demonstrate, contextualized tasks reduce nuanced hallucinations, and Huang et al. (2025) show that structured, ethically constrained prompts help mitigate bias in large language models.”
Foundation Models for Education: Promises and Prospects
Xu, T., Tong, R. J., Liang, J., Fan, X., Li, H., & Wen, Q. (2024) · IEEE Intelligent Systems · doi:10.1109/mis.2024.3398191 · Cited by 30
“However, concerns arise regarding the rigor of foundation models, as some early applications have shown XXXX-XXX © 2024 IEEE Digital Object Identifier 10.1109/XXX.0000.0000000 limitations 7 .”
BackgroundInfluentialPBChat: Enhance Student’s Problem Behavior Diagnosis with Large Language Model
Chen, P., Fan, Z., Lu, Y., & Xu, Q. M. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-64302-6_3 · Cited by 12
A Chat Bot for Enrollment of Xi 'an Jiaotong-Liverpool University Based on RAG*
Xu, L., & Liu, J. (2024) · International Workshop on Advanced Algorithms and Control Engineering (IWAACE) · doi:10.1109/iwceaa63616.2024.10823979 · Cited by 11
Neupane, S., Tripathi, H., Mitra, S., Bozorgzad, S., Mittal, S., Rahimi, S., & Amirlatifi, A. (2024) · IEEE International Conference on Big Data (BigData) · doi:10.1109/bigdata62323.2024.10825266 · Cited by 19
Reducing Emissions Using Artificial Intelligence in the Energy Sector: A Scoping Review
Alatalo, J., Heilimo, E., Rantonen, M., Väänänen, O., & Sipola, T. (2025) · Applied Sciences · doi:10.3390/app15020999 · Cited by 13
“Hallucinations are problematic when using generative AI to answer questions [46].”
Antibiotics and Artificial Intelligence: Clinical Considerations on a Rapidly Evolving Landscape
Giacobbe, D. R., Guastavino, S., Marelli, C., Murgia, Y., Mora, S., Signori, A., Rosso, N., Giacomini, M. L., Campi, C., Piana, M., et al. (2025) · Infectious Diseases and Therapy · doi:10.1007/s40121-025-01114-5 · Cited by 13
How to Write Effective Prompts for Screening Biomedical Literature Using Large Language Models
Colangelo, M. T., Guizzardi, S., Meleti, M., Calciolari, E., & Galli, C. (2025) · BioMedInformatics · doi:10.3390/biomedinformatics5010015 · Cited by 14
“One notable issue involves hallucinations, where the model confidently supplies factually incorrect information [58].”
On the disagreement problem in Human-in-the-Loop federated machine learning
Huelser, M. J. M., Mueller, H., Díaz-Rodríguez, N., & Holzinger, A. (2025) · Journal of Industrial Information Integration · doi:10.1016/j.jii.2025.100827 · Cited by 15
NumScout: Unveiling Numerical Defects in Smart Contracts Using LLM-Pruning Symbolic Execution
Chen, J., Shao, Z., Yang, S., Shen, Y., Wang, Y., Chen, T., Shan, Z., & Zheng, Z. (2025) · IEEE Transactions on Software Engineering · doi:10.1109/tse.2025.3555622 · Cited by 17
“This allows us to analyze the original logical structure and meaning of expressions during symbolic execution.”
Generative Agents in Agent-Based Modeling: Overview, Validation, and Emerging Challenges
Adornetto, C., Mora, A., Hu, K., Garcia, L. I., Atchade-Adelomou, P., Greco, G., Alonso, L., & Larson, K. (2025) · IEEE Transactions on Artificial Intelligence · doi:10.1109/tai.2025.3566362 · Cited by 38
“Hallucinations in LLMs refer to generating factually incorrect or ungrounded outputs, categorized into factuality and faithfulness hallucinations [73].”
He, Y., Zhu, X., Li, D., & Wang, H. (2025) · Electronics · doi:10.3390/electronics14101961 · Cited by 28
Self-citationHeterRAG: Heterogeneous Processing-in-Memory Acceleration for Retrieval-augmented Generation
Liu, C., Liu, H., Chen, D., Huang, Y., Zhang, Y., Xiao, W., Liao, X., & Jin, H. (2025) · Proceedings - International Symposium on Computer Architecture/Proceedings · doi:10.1145/3695053.3731089 · Cited by 26
MedKA: A knowledge graph-augmented approach to improve factuality in medical Large Language Models
Deng, Y., Zhao, S., Miao, Y., Zhu, J., & Li, J. (2025) · Journal of Biomedical Informatics · doi:10.1016/j.jbi.2025.104871 · Cited by 24
Seeing Far and Clearly: Mitigating Hallucinations in MLLMs with Attention Causal Decoding
Tang, F., Liu, C., Xu, Z., Hu, M., Huang, Z., Xue, H., Chen, Z., Peng, Z., Yang, Z., Zhou, S., et al. (2025) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52734.2025.02435 · Cited by 68
“Hallucination may originate from reliance on model priors [7, 19, 26, 33, 39, 43, 86, 93], limited knowledge comprehension [21, 42, 56, 80, 96], or an inability to effectively contextualize the given input [8, 22, 36, 48, 49, 69, 76].”
Boretti, A. (2026) · Energy Research & Social Science · doi:10.1016/j.erss.2026.104720 · Cited by 10
A Lifecycle-Oriented Survey of Emerging Threats and Vulnerabilities in Large Language Models
de Maio, C., Di Gisi, M., Fenza, G., Gallo, M., & Loia, V. (2025) · IEEE Access · doi:10.1109/access.2025.3619764 · Cited by 10
“Surveys such as [17] provide a structured view of their causes, detection methods, and mitigation strategies, while complementary taxonomies (e.g., [18]) emphasize the diverse manifestations of hallucination, from factual to logical inconsistencies, each carrying distinct risks in sensitive domains…”
Multi-Layered Framework for LLM Hallucination Mitigation in High-Stakes Applications: A Tutorial
Hiriyanna, S., & Zhao, W. (2025) · Computers · doi:10.3390/computers14080332 · Cited by 21
“Rather than truly understanding concepts, they excel at pattern matching and statistical prediction based on vast amounts of training data [4].”
Foundation Models for Autonomous Driving Perception: A Survey Through Core Capabilities
Sathyam, R., & Li, Y. (2025) · IEEE Open Journal of Vehicular Technology · doi:10.1109/ojvt.2025.3604823 · Cited by 19
“Hallucination Risk : The phenomenon of hallucination— where models produce outputs inconsistent with the provided sensor reality—poses a severe safety risk [61].”
Vach, M., Gliem, M., Weiss, D., Ivan, V. L., Hauke, F., Boschenriedter, C., Rubbert, C., & Caspers, J. (2025) · Clinical Neuroradiology · doi:10.1007/s00062-025-01562-z · Cited by 9
“However, a critical challenge with LLMs is their tendency to fabricate facts—a phenomenon called “hallucina-tion”—which is particularly concerning in medical contexts [6].”
Modeling semantic representation with LLM-enhanced for knowledge-aware recommendation
Liu, J., Yi, B., Zhang, H., Shen, X., Song, L., Yu, L., & Zheng, H. (2025) · Information Processing & Management · doi:10.1016/j.ipm.2025.104387 · Cited by 11
Masanneck, L., Epping, P. Z., Meuth, S. G., & Pawlitzki, M. (2025) · Journal of Medical Internet Research · doi:10.2196/79379 · Cited by 13
“Despite these promising applications, LLMs often struggle with factual accuracy and reasoning and are known to “hallucinate” [6,7], which implies that they can produce fluent, seemingly authoritative responses that are entirely fabricated or factually incorrect.”
Guided and knowledgeable multi-agent debate for fact verification
Ma, X., Rao, G., Xu, L., Wang, X., Fan, Z., & Zhang, Z. (2025) · Expert Systems with Applications · doi:10.1016/j.eswa.2025.130103 · Cited by 13
Xu, W., Wan, H., Goel, S., & Antonopoulos, C. (2025) · Energy and Buildings · doi:10.1016/j.enbuild.2025.116712 · Cited by 18
Wei, C., Zhou, L., Liang, B., Chen, J., Wang, G., & Li, X. (2025) · Food Chemistry · doi:10.1016/j.foodchem.2025.147377 · Cited by 11
Liu, J., Yu, W., Yang, M., & Liu, F. (2025) · Safety Science · doi:10.1016/j.ssci.2025.107090 · Cited by 12
Self-citationKnowledge graph and large language model synergy for food safety: Approaches and perspectives
An, H., Dong, Q., Doulgeraki, A. I., Nychas, G.-J. E., Zhou, Z., & Liu, Y. (2026) · Trends in Food Science & Technology · doi:10.1016/j.tifs.2026.105592 · Cited by 14
Kamińska, A. M. (2026) · The Journal of Academic Librarianship · doi:10.1016/j.acalib.2026.103218 · Cited by 12
LLLMs: A Data-Driven Survey of Evolving Research on Limitations of Large Language Models
Kostikova, A., Wang, Z., Bajri, D., Pütz, O., Paaßen, B., & Eger, S. (2026) · ACM Computing Surveys · doi:10.1145/3801096 · Cited by 38
Hodzic, S., Stević, A., & Matthes, J. (2026) · Technology in Society · doi:10.1016/j.techsoc.2026.103331 · Cited by 12
Consistent Game Content Creation via Function Calling for Large Language Models
Gallotta, R., Liapis, A., & Yannakakis, G. N. (2024) · IEEE Conference on Games (CoG) · doi:10.1109/cog60054.2024.10645599 · Cited by 18
“Following [7], we identify context inconsistency , where parts of the context provided to the LLM is ignored, as the class of hallucinations we are most concerned with, and evaluate our approach against it by defining our own model accuracy measure.”
MethodRetrieval Augmented Generation on Hybrid Cloud: A New Architecture for Knowledge Base Systems
Chuang, C., & Chen, K.-C. (2024) · IIAI International Congress on Advanced Applied Informatics (IIAI-AAI) · doi:10.1109/iiai-aai63651.2024.00022 · Cited by 13
“One of the most notable issues is “hallucinations” [1], which the model generates information that is not correct.”
BackgroundAn Interactive Multi-Modal Query Answering System with Retrieval-Augmented Large Language Models
Wang, M., Wu, H., Ke, X., Gao, Y., Xu, X., & Chen, L. (2024) · Proceedings of the VLDB Endowment · doi:10.14778/3685800.3685868 · Cited by 21
A Causal Explainable Guardrails for Large Language Models
Chu, Z., Wang, Y., Li, L., Wang, Z., Qin, Z., & Ren, K. (2024) · ACM SIGSAC Conference on Computer and Communications Security (CCS) · doi:10.1145/3658644.3690217 · Cited by 42
Cinquin, O. (2024) · Briefings in Bioinformatics · doi:10.1093/bib/bbaf045 · Cited by 12
Tian, W., Chen, R., & Chen, L. (2025) · Analytical Chemistry · doi:10.1021/acs.analchem.4c07047 · Cited by 13
Bombieri, M., Fiorini, P., Ponzetto, S. P., & Rospocher, M. (2025) · ACM Transactions on Intelligent Systems and Technology · doi:10.1145/3725852 · Cited by 18
Enhancing Scene Graph Generation with Hierarchical Relationships and Commonsense Knowledge
Jiang, B., Zhuang, Z., Shivakumar, S. S., & Taylor, C. J. (2025) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv61041.2025.00861 · Cited by 28
“Statistics Since both S aligned and S violated are collected purely from the training dataset, we address concerns about data distribution mismatches by showing some statistics in Table 6.”
Zhu, Z., Liu, J., Hong, C. W., Houshmand, S., Wang, K., & Yang, Y. (2025) · American Journal of Roentgenology · doi:10.2214/ajr.25.32729 · Cited by 12
Cognitive Computing with Large Language Models for Student Assessment Feedback
Abbas, N., & Atwell, E. (2025) · Big Data and Cognitive Computing · doi:10.3390/bdcc9050112 · Cited by 15
Simulating Oxford-Style Debates with LLM-Based Multi-Agent Systems
Harbar, Y., & Chudziak, J. A. (2025) · Lecture notes in computer science · doi:10.1007/978-981-96-6008-7_21 · Cited by 20
IoT Rule Generation With Cross-View Contrastive Learning and Perplexity-Based Ranking
Cimino, G., Deufemia, V., & Limone, M. (2025) · IEEE Internet of Things Journal · doi:10.1109/jiot.2025.3579920 · Cited by 10
“AutoIoT and TAPChecker provide LLMs with device information in the input but lack rule format validation, while ChatIoT uses an LLM for format validation, risking hallucinations [36] that may misrepresent rule components.”
Detecting and Reducing the Factual Hallucinations of Large Language Models with Metamorphic Testing
Wu, W., Cao, Y., Yi, N., Ou, R., & Zheng, Z. (2025) · Proceedings of the ACM on software engineering. · doi:10.1145/3715784 · Cited by 22
In-Storage Acceleration of Retrieval Augmented Generation as a Service
Mahapatra, R., Santhanam, H., Priebe, C., Xu, H., & Esmaeilzadeh, H. (2025) · Proceedings - International Symposium on Computer Architecture/Proceedings · doi:10.1145/3695053.3731032 · Cited by 24
Ahn, J., Kang, B. G., Chang, M. Y., & Yoon, S. (2025) · Clinical and Experimental Otorhinolaryngology · doi:10.21053/ceo.2025-00121 · Cited by 10
“Even LLMs trained with RLHF may encounter challenges such as hallucinations [35], in which they produce plausible-sounding but factually incorrect information [36].”
Fu, J., Lu, Y., Yang, Z., Nah, F. F., & Lc, R. (2025) · Designing Interactive Systems Conference · doi:10.1145/3715336.3735812 · Cited by 17
AI Veterinary Assistance: Enhancing Clinical Decision-Making in Animal Healthcare
Jin, Y.-G., Wu, G., Seo, J. Y., Park, S.-J., Hur, S.-H., Aliyeva, D., Park, J.-H., & Kim, K.-M. (2025) · IEEE Access · doi:10.1109/access.2025.3587787 · Cited by 14
“Even if a model’s response exhibits a natural logical flow, independent verification remains necessary to ensure alignment with clinical standards [69].”
Madrid-García, A., Benavent, D., & Merino-Barbancho, B. (2025) · EULAR Rheumatology Open · doi:10.1016/j.ero.2025.06.012 · Cited by 9
“For clarity, a de fi nition box summarising key terms is shown in Table 1 [4,6,17,18].”
Artificial intelligence in web accessibility: A systematic mapping study
Campoverde‐Molina, M., & Luján‐Mora, S. (2025) · Computer Standards & Interfaces · doi:10.1016/j.csi.2025.104055 · Cited by 19
MAO: A Framework for Process Model Generation With Multi-Agent Orchestration
Lin, L., Jin, Y., Zhou, Y., Chen, W., & Qian, C. (2025) · IEEE Transactions on Services Computing · doi:10.1109/tsc.2025.3611816 · Cited by 34
“In nat-ural language processing, “hallucination” denotes the situation where a machine learning model generates text or responses that are entirely disconnected from the context or not reliant on the input data [52].”
How to Use Generative AI in Educational Research
Roe, J. (2025) · Cambridge University Press eBooks · doi:10.1017/9781009675338 · Cited by 22
Ontology-Aware RAG for Improved Question-Answering in Cybersecurity Education
Zhao, C., Agrawal, G., Zhang, F., Kumarage, T., Tan, Z., Deng, Y., Chen, Y.-C., & Liu, H. (2025) · IEEE International Conference on Big Data (BigData) · doi:10.1109/bigdata66926.2025.11401740 · Cited by 41
“While LLM-powered QA systems hold great promise for enhancing learning, they also face challenges such as hallucination and limited domain knowledge, which can undermine their effectiveness [9], [10].”
GhostCite: A Large-Scale Analysis of Citation Validity in the Age of Large Language Models
Xu, Z., Qiu, Y., Sun, L., Miao, F., Wu, F., Li, X., Wang, X., Lu, H., Zhang, Z., Hu, Y., et al. (2026) · arXiv · doi:10.48550/arxiv.2602.06718 · Cited by 3
“The phenomenon of LLM hallucination has been extensively studied [24]–[26], with factuality benchmarks such as TruthfulQA [27] and HaluEval [28] quantifying these failures.”
Delmas, M., Wysocka, M., & Freitas, A. (2024) · Computational Linguistics · doi:10.1162/coli_a_00520 · Cited by 17
Learner Modeling Interpretability and Explainability in Intelligent Adaptive Systems
Zapata‐Rivera, D., & Arslan, B. (2024) · Integrated science · doi:10.1007/978-3-031-58363-6_7 · Cited by 10
Hallu-PI: Evaluating Hallucination in Multi-modal Large Language Models within Perturbed Inputs
Ding, P., Wu, J., Kuang, J., Ma, D., Cao, X., Cai, X., Chen, S., Chen, J., & Huang, S. (2024) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3664647.3681251 · Cited by 31
Aligning large language models and geometric deep models for protein representation
Shu, D., Duan, B., Guo, K. Z., Zhou, K., Tang, J., & Du, M. (2025) · Patterns · doi:10.1016/j.patter.2025.101227 · Cited by 14
“26,27 To evaluate the impact of alignment on both performance and hallucination, we selected the protein description task and employed ROUGE 28 and BLEU 29 metrics.”
Huang, L., Lin, A., Li, H., Wang, Q., Shen, J., Jiang, A., Chang, Q., Gan, W., Zhu, L., Mou, W., et al. (2025) · View · doi:10.1002/viw.20250077 · Cited by 9
Self-citationDAWN: Designing Distributed Agents in a Worldwide Network
Aminiranjbar, Z., Tang, J., Wang, Q., Pant, S., & Viswanathan, M. (2025) · IEEE Access · doi:10.1109/access.2025.3588425 · Cited by 19
“First, there are vulnerabilities inherent to LLMs, such as hallucinations [56], attacks that leverage tuned instructional methods like jailbreaking [57], [58], and prompt injection [59].”
BackgroundGün, M. (2025) · BMC Emergency Medicine · doi:10.1186/s12873-025-01303-y · Cited by 8
“…as hallucinations, several critical errors—such as recommending thrombolysis in a stroke case with unknown onset or failing to perform essential acid-base calculations like the anion gap or delta ratio—may reflect hallucination-like behavior or underlying limitations in reasoning capacity [22, 23].”
A Survey of Human Intelligence Augmented Artificial Intelligence: An Autonomous Driving Perspective
Li, S., Yuan, K., Tang, Z., Zhang, Y., Ji, P., Ge, Q., Du, S., Huang, Y., & Chen, H. (2025) · Automotive Innovation · doi:10.1007/s42154-025-00361-z · Cited by 13
Fine-tuning LLMs in behavioral psychology for scalable health coaching
Mantena, S., Johnson, A., Oppezzo, M., Schütz, N., Tolas, A., Doijad, R., Mattsson, C. M., Lawrie, A., Ramírez‐Posada, M., Schmiedmayer, P., et al. (2025) · npj Cardiovascular Health · doi:10.1038/s44325-025-00083-5 · Cited by 24
“This represents an important area for future work, especially as models begin to incorporate patient-speci fi c information or deliver more targeted health advice 68,69 .”
ProveRAG: Provenance-Driven Vulnerability Analysis With Automated Retrieval-Augmented LLMs
Fayyazi, R., Trueba, S. H., Zuzak, M., & Yang, S. J. (2025) · IEEE Access · doi:10.1109/access.2025.3638251 · Cited by 27
“However, prior work has identified three aspects that limit the broader adoption of LLMs to cybersecurity applications [8, 16, 28, 37], which we aim to address in this work.”
Zhao, Y., Zhou, J., Bi, D., Mihalj, T., Hu, J., & Eichberger, A. (2026) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2026.3651004 · Cited by 32
“Huang et al. [243] categorize countermeasures into three levels: data, training, and inference and identify RAG as a promising approach.”
Large language models for spreading dynamics in complex systems
Jiang, S., Ren, H., Gao, Y., Zhang, Y., Li, Q., Xiao, D., Fan, J., Tang, R., & Wang, W. (2026) · Physics Reports · doi:10.1016/j.physrep.2026.01.006 · Cited by 11
DAWN: Designing Distributed Agents in a Worldwide Network
Aminiranjbar, Z., Tang, J., Wang, Q., Pant, S., & Viswanathan, M. (2024) · arXiv · doi:10.48550/arxiv.2410.22339 · Cited by 8
Miao, Y., Wen, J., Luo, Y., & Li, J. (2026) · International Journal of Medical Informatics · doi:10.1016/j.ijmedinf.2025.106136 · Cited by 10
Kaiser, C., Kaiser, J., Schallner, R., & Schneider, S. (2025) · Proceedings of the Extended Abstracts of the CHI Conference on Human Factors in Computing Systems · doi:10.1145/3706599.3720123 · Cited by 11
Large Language Models for the Automated Analysis of Optimization Algorithms
Sartori, C. C., Blum, C., & Ochoa, G. (2024) · Genetic and Evolutionary Computation Conference (GECCO) · doi:10.1145/3638529.3654086 · Cited by 16
Liu, N., Wei, K., Yang, Y., Tao, J., Sun, X., Yao, F., Yu, H., Jin, L., Lv, Z., & Fan, C. (2024) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2024.3447778 · Cited by 19
“Although the overall summaries are fluent, a few generated results exhibit hallucinations [72], i.e., generating unsupported or fabricated content, as shown in Table XIV.”
BackgroundLiu, D., Lei, X., Huang, Y. Z., Liu, W., & Zhang, J. (2024) · Journal of Asian Architecture and Building Engineering · doi:10.1080/13467581.2024.2402771 · Cited by 10
“(Huang et al. 2023).”
Benchmarking large language models on safety risks in scientific laboratories
Zhou, Y., Yang, J., Huang, Y., Guo, K., Emory, Z. C., Ghosh, B., Bedar, A., Shekar, S., Liang, Z., Chen, P., et al. (2026) · Nature Machine Intelligence · doi:10.1038/s42256-025-01152-1 · Cited by 54
“However, these advanced capabilities also introduce substantial risks when hallucinations 23 , delayed hazard detection or protocol misinterpretations intersect with physical Article https://doi.org/10.1038/s42256-025-01152-1 Specifically, we aim to answer the following key research questions: can…”
InfluentialYuan, Z., Xie, F., & Ji, T. (2024) · International Conference on Robotics and Computer Vision (ICRCV) · doi:10.1109/icrcv62709.2024.10758606 · Cited by 11
“This severe LLM hallucination causes the agent to repeatedly believe it is in danger [26], [27], constantly switching between stopping and moving forward, greatly slowing down its speed.”
Multimodal LLMs Can Reason about Aesthetics in Zero-Shot
Jiang, R., & Chen, C. W. (2025) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3746027.3754961 · Cited by 30
Kuan, C.-Y., & Lee, H.-Y. (2025) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp49660.2025.10888384 · Cited by 50
Oladokun, B. D., Enakrire, R. T., Emmanuel, A. K., Ajani, Y. A., & Adetayo, A. J. (2025) · Journal of Web Librarianship · doi:10.1080/19322909.2025.2482093 · Cited by 7
“Also, Huang et al. (2023) provided a taxonomy and discussion of hallucinations in large language models, emphasizing the need for nuanced understanding and advancements in addressing these challenges.”
Iatrellis, O., Bania, A., Samaras, N., Kosmopoulou, I., & Panagiotakopoulos, T. (2025) · International journal of doctoral studies · doi:10.28945/5579 · Cited by 15
“Scholars have increasingly voiced concerns regarding AI hallucination instances where AI confidently provides incorrect or misleading information, as well as inherent biases in AI outputs that can reinforce stereotypes or skew perspectives (Bang et al., 2023; Huang et al., 2025).”
Αn AI-driven approach to assess sentiments and interpret context in a critical mineral supply chain
Joglekar, N. R., Nie, W., Fonseca, M. A., Tsolakis, N., & Kumar, M. (2025) · International Journal of Production Research · doi:10.1080/00207543.2025.2532142 · Cited by 12
AlSammarraie, A., Al-Saifi, A., Kamhia, H., Aboagla, M., & Househ, M. (2025) · BMJ Health & Care Informatics · doi:10.1136/bmjhci-2025-101570 · Cited by 10
LUK: Empowering Log Understanding With Expert Knowledge From Large Language Models
Ma, L., Yang, W., Jiang, S., Fei, B., Zhou, M., Li, S., Zhao, M., Xu, B., & Xiao, Y. (2025) · IEEE Transactions on Software Engineering · doi:10.1109/tse.2025.3594046 · Cited by 17
“The BGP instance name is provided in the log (e.g. , [InstanceName]).”
Multilingual graph retrieval-augmented generation for product design using design knowledge
Zhang, H., Wang, T., Liang, Z., Huang, Z., Chen, C., & Cheng, L. (2025) · Journal of Engineering Design · doi:10.1080/09544828.2025.2537464 · Cited by 15
Using LLMs for self-care: User and counsellor perspectives
Wester, J., de Jong, S., Pohl, H., & van Berkel, N. (2025) · International Journal of Human-Computer Studies · doi:10.1016/j.ijhcs.2025.103589 · Cited by 11
Mastrianni, A., Twede, H., Sarcevic, A., Wander, J. D., Austin-Tse, C. A., Saponas, S., Rehm, H. L., Conard, A. M., & Hall, A. K. (2025) · ACM Transactions on Interactive Intelligent Systems · doi:10.1145/3756326 · Cited by 14
A Zero-Training Error Correction System with Large Language Models
Wu, Y., Yang, C., Zhu, M., Miao, X., Ni, W., Xi, M., Zhao, X., & Yin, J. (2025) · IEEE International Conference on Data Engineering (ICDE) · doi:10.1109/icde65448.2025.00221 · Cited by 11
“Second , unique and non-standard data errors can induce LLMs to generate plausible but incorrect or fabricated corrections, a phenomenon known as “hallucination” [29], [30] ( CH2 ).”
Liu, H., Chen, Z., Li, P., Liu, Y.-Z., Liu, X., Xu, R. X., & Sun, M. (2025) · Journal of Biomedical Informatics · doi:10.1016/j.jbi.2025.104896 · Cited by 11
AutoPT: How Far Are We From the Fully Automated Web Penetration Testing?
Wu, B., Chen, G., Chen, K., Shang, X., Han, J., He, Y., Zhang, W., & Yu, N. (2025) · IEEE Transactions on Information Forensics and Security · doi:10.1109/tifs.2025.3601343 · Cited by 50
“Third, consistent with previous work [28], LLM has problems with inaccurate result generation and hallucinations in reasoning, which is more severe for agents.”
Word2Wave: Language Driven Mission Programming for Efficient Subsea Deployments of Marine Robots
Chen, R., Blow, D., Abdullah, A., & Islam, M. J. (2025) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra55743.2025.11128181 · Cited by 22
“This is a common issue with language models where overgeneralization outputs plausible but non-functional commands [75].”
The AI Annotator: Large Language Models’ Potential in Scoring Sustainability Reports
Wu, Y., Hu, P., & Wang, D. D. (2025) · Systems · doi:10.3390/systems13100899 · Cited by 12
“According to Huang et al., hallucinations are classified into two main categories: factual hallucinations and faithful hallucinations [78].”
Metadata
- Authors
- Huang, L., Yu, W., Ma, W., Zhong, W., Feng, Z., Wang, H., Chen, Q., Peng, W., Feng, X., Qin, B., & Liu, T.
- Affiliations
- Harbin Institute of Technology (CN)
- Huawei Technologies (China) (CN)
- ORCID
- Venue
- ACM Transactions on Information Systems
- Published
- 2025-01-24
- Volume
- 43
- Issue
- 2
- Pages
- 1-55
- ISSN
- 1046-8188 (ISSN-L)
- 1558-2868
- OpenAlex
- W4404534210
- Semantic Scholar
- 1e909e2a8cdacdcdff125ebcc566f37cb869a1c8
- Cited by
- 4,688
- References
- 229
- Citations by year
- 2026: 2433 · 2025: 1643
- FWCI
- 504.87
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- ACM Transactions on Information Systems (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Physical Sciences › Computer Science › Information Systems › Information Retrieval and Search Behavior
- Physical Sciences › Computer Science › Artificial Intelligence › Natural Language Processing Techniques
- Keywords
- LLM hallucination, large language models, hallucination taxonomy, hallucination detection, hallucination mitigation, retrieval-augmented generation, information retrieval systems, factual reliability, large vision-language models, knowledge boundaries
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-27, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-27 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.