---
title: "HalluTruthQA — paper by Ghaly; and Hadid (2026)"
description: "Arabic LLM answers can read well and still contain factual errors. 28 references, cited by 3 papers here."
authors: ["Abdessalam Bouchekif; Mohammed-En-Nadhir Zighem; Salah Eddine Bekhouche; Hichem Telli; Somaya Eltanbouly; Shahd Gaben; Heba Sbahi; Samer Rashwani; Mutaz Al-Khatib; Emad Mohamed; Mohammed Ghaly;", "Abdenour Hadid"]
year: 2026
venue: "arXiv:2607.20219"
source: "https://arxiv.org/abs/2607.20219"
doi: "10.48550/arxiv.2607.20219"
arxiv: "2607.20219"
licence: "CC BY-NC-ND 4.0 (arXiv)"
licence_url: "http://creativecommons.org/licenses/by-nc-nd/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/bouchekif-2026-hallutruthqa-fine-grained-benchmark-hallucination/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/bouchekif-2026-hallutruthqa-fine-grained-benchmark-hallucination/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/5/) / HalluTruthQA: A Fine-Grained Benchmark for Hallucination Detection, Localization, and Explanation in Arabic Question Answering

# HalluTruthQA: A Fine-Grained Benchmark for Hallucination Detection, Localization, and Explanation in Arabic Question Answering

**Authors:** Abdessalam Bouchekif; Mohammed-En-Nadhir Zighem; Salah Eddine Bekhouche; Hichem Telli; Somaya Eltanbouly; Shahd Gaben; Heba Sbahi; Samer Rashwani; Mutaz Al-Khatib; Emad Mohamed; Mohammed Ghaly; and [Abdenour Hadid](https://md.research.elamur.ai/free/en/authors/abdenour-hadid/) · **Year:** 2026 · **Venue:** arXiv:2607.20219 · **Licence:** [CC BY-NC-ND 4.0 (arXiv)](http://creativecommons.org/licenses/by-nc-nd/4.0/)

**Links:** [Publisher's page](https://arxiv.org/abs/2607.20219) · [PDF](https://arxiv.org/pdf/2607.20219) · [DOI 10.48550/arxiv.2607.20219](https://doi.org/10.48550/arxiv.2607.20219) · [arXiv 2607.20219](https://arxiv.org/abs/2607.20219) · [OpenAlex](https://openalex.org/W7170180213)

**Citations:** 0 citations (OpenAlex, 2026-10-14) · [3](https://md.research.elamur.ai/free/en/papers/bouchekif-2026-hallutruthqa-fine-grained-benchmark-hallucination/cited-by/) in our library, of the 149,439 papers whose bibliographies we read

## Summary
Arabic LLM answers can read well and still contain factual errors. Most existing benchmarks only say whether a whole answer is right or wrong, so they cannot show where the mistake is, why it is a mistake, or whether a model could still pick the right fact. HalluTruthQA fills this gap with 2,400 expert-curated Arabic QA items in four domains: Islamic knowledge, history, science and geography. All answers come from a single generator, Fanar-1-9B-Instruct. Each item has:
- a checked correct answer and a yes/no label marking whether the response is hallucinated;
- six hand-written candidate answers for multiple-choice verification;
- for hallucinated answers, character-level error spans, a human explanation, and macro- and micro-level error types.

Four open models (ALLaM-7B, Falcon-H1R-7B, Qwen3-32B, SILMA) were tested zero-shot on four tasks: detection, span localization, factual verification and explanation. Fanar was also tested on its own outputs as a separate reference. No model was best at everything, which suggests the four tasks measure different skills. The benchmark therefore argues for error analysis that goes beyond a single yes/no label, especially in source-sensitive domains.

## Key points
- 1,013 of the 2,400 answers (42.2%) are hallucinated, with 1,120 annotated error spans in total. The Islamic domain has the highest rate (56.8%) and history the lowest (32.2%).
- Best scores:

  | Task | Metric | Best score |
  |---|---|---|
  | Detection | Macro-F1 | 0.880 |
  | Span localization | F1-Sp | 0.516 |
  | Factual verification | LO-Score | 0.852 |
  | Explanation | — | 0.644 |

- Different systems lead on different tasks:
  - Qwen3-32B localizes spans best.
  - ALLaM-7B gives the most reliable explanations.
  - FANAR-9B does well when judging its own outputs.
- 39 items fooled all five systems in the same way. About 79.5% of these were valid answers wrongly flagged as hallucinated, and history accounted for the most of them (18).
- Error patterns differ by domain:
  - In Islamic content, mistakes often concern evidence and source attribution (Qur'anic references, hadith attributions).
  - In the other domains, they are mostly contradictions or invented details.
  - Explanations are scored by an LLM judge, and a human-checked sample is used to validate the judge.

## Interesting ideas
- The five distractors per question are written to look like the generator's own output, so a model cannot find the right option from surface style alone.
- Each error span gets two labels: a broad failure class (contradiction, fabrication, context inconsistency) and a specific error kind (entity, date, number, citation, attribution).
- The paper keeps the main short answer apart from its supporting justification, because an answer can be correct while its reasoning is invented.
- The authors study items where every system fails in the same way and find most of them are false alarms caused by paraphrases.
- When the generator judges its own outputs, those results are reported apart from the independent evaluators.

## For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

- **I1** (answer): Distractors for fable and school-test QA could be produced in the same voice as the correct answer, so readers are rewarded for meaning rather than wording cues.
- **I2** (back): Mismatches in the text → graph → text compressor could be labeled the same way, showing whether losses come from entities, quantities or time.
- **I3** (world): Every supporting claim in an answer could be checked against the world of the text separately from the core answer, and unsupported side claims could be flagged.
- **I4** (learn): Items where all witness parsers agree but are wrong should be mined as a separate negative-control set, since agreement-based silver data cannot catch them.
- **I5** (learn): Self-checks by the same LLM teacher tier that produced a label should be scored separately from independent checks, so self-agreement does not inflate quality estimates.

*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

- [Hamad bin Khalifa University](https://md.research.elamur.ai/free/en/universities/hamad-bin-khalifa-university/) — University, Doha, Qatar · 141 works · 14,175 citations
- [University of the Basque Country](https://md.research.elamur.ai/free/en/universities/university-of-the-basque-country/) — University, Leioa, Spain · 57 works · 16,520 citations
- [Nazarbayev University](https://md.research.elamur.ai/free/en/universities/nazarbayev-university/) — University, Astana, Kazakhstan · 4 works · 44 citations
- [University of Biskra](https://md.research.elamur.ai/free/en/universities/university-of-biskra/) — University, Biskra, Algeria · 4 works · 7 citations
- [Universiti Malaysia Kelantan](https://md.research.elamur.ai/free/en/universities/universiti-malaysia-kelantan/) — University, Kota Bharu, Malaysia · 3 works · 4 citations

📚 Shelves (5): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 16,194 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 16,039 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 15,891 works · [👁️ Computer Vision and Pattern Recognition](https://md.research.elamur.ai/free/en/shelves/computer-science/computer-vision-and-pattern-recognition/) · 2,869 works

🧭 Topics (3): [Semitic languages](https://md.research.elamur.ai/free/en/topics/semitic-languages/) · 2,908 works · [📏 Evaluation & benchmarks](https://md.research.elamur.ai/free/en/topics/evaluation-benchmarks/) · 1,220 works · [❓ Question answering](https://md.research.elamur.ai/free/en/topics/question-answering/) · 441 works

## Cite

Cite as: Abdessalam Bouchekif; Mohammed-En-Nadhir Zighem; Salah Eddine Bekhouche; Hichem Telli; Somaya Eltanbouly; Shahd Gaben; Heba Sbahi; Samer Rashwani; Mutaz Al-Khatib; Emad Mohamed; Mohammed Ghaly; and Abdenour Hadid. 2026. HalluTruthQA: A Fine-Grained Benchmark for Hallucination Detection, Localization, and Explanation in Arabic Question Answering. arXiv:2607.20219.

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@misc{ghaly2026hallutruthqa,
  title = {{HalluTruthQA: A Fine-Grained Benchmark for Hallucination Detection, Localization, and Explanation in Arabic Question Answering}},
  author = {Abdessalam Bouchekif; Mohammed-En-Nadhir Zighem; Salah Eddine Bekhouche; Hichem Telli; Somaya Eltanbouly; Shahd Gaben; Heba Sbahi; Samer Rashwani; Mutaz Al-Khatib; Emad Mohamed; Mohammed Ghaly; and Abdenour Hadid},
  year = {2026},
  doi = {10.48550/arxiv.2607.20219},
  eprint = {2607.20219},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2607.20219}
}
```

- [← previous HalluScore: Large Language Model Hallucination Question Answering Benchmark](https://md.research.elamur.ai/free/en/papers/alansari-2026-halluscore-large-language-model-hallucination/)
- [next → Harnessing Negative Signals: Reinforcement Distillation from Teacher Data for LLM Reasoning](https://md.research.elamur.ai/free/en/papers/xu-2026-harnessing-negative-signals-reinforcement-distillation/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/5/)

## References

28 references, 15 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Samir Abdaljalil, Hasan Kurban, and Erchin Serpedin. 2025. Halluverse25: Fine-grained multilingual benchmark dataset for llm hallucinations. arXiv preprint arXiv:2503.07833. [↗](https://arxiv.org/abs/2503.07833)
2. Samir Abdaljalil, Parichit Sharma, Erchin Serpedin, and Hasan Kurban. 2026. Halluverse-m^ 3: A multitask multilingual benchmark for hallucination in llms. arXiv preprint arXiv:2602.06920. [↗](https://arxiv.org/abs/2602.06920)
3. Aisha Alansari and Hamzah Luqman. 2025. Ara-HalluEval: A fine-grained hallucination evaluation framework for Arabic LLMs. In Proceedings of The Third Arabic Natural Language Processing Conference, pages 148–161, Suzhou, China. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Aisha+Alansari+and+Hamzah+Luqman%2E+2025%2E+Ara%2DHalluEval%3A+A+fine%2Dgrained+hallucination+evaluation+framework+for+Arabic+LLMs%2E+In+Proceedings+of+The+Third+Arabic+Natural+Language+Processing+Conference%2C+pag)
4. Aisha Alansari and Hamzah Luqman. 2026a. [Halluscore: Large language model hallucination question answering benchmark](https://md.research.elamur.ai/free/en/papers/alansari-2026-halluscore-large-language-model-hallucination/). arXiv preprint arXiv:2605.17007. [↗](https://arxiv.org/abs/2605.17007)
5. Aisha Alansari and Hamzah Luqman. 2026b. Large language models hallucination: A comprehensive survey. arXiv preprint arXiv:2510.06265. [↗](https://arxiv.org/abs/2510.06265)
6. Fakhraddin Alwajih, Abdellah El Mekki, Hamdy Mubarak, Majd Hawasly, Abubakr Mohamed, and Muhammad Abdul-Mageed. 2025. [Palmx 2025: The first shared task on benchmarking llms on arabic and islamic culture](https://md.research.elamur.ai/free/en/papers/alwajih-2025-palmx-2025-first-shared-task/). In Proceedings of The Third Arabic Natural Language Processing Conference: Shared Tasks, pages 774– 789.
7. Abdessalam Bouchekif, Somaya Eltanbouly, Samer Rashwani, Shahd Gaben, Mutaz Al-Khatib, Heba Sbahi, Emad Mohamed, and Mohammed Ghaly. 2026a. [Qias 2026: Overview of the shared task on islamic inheritance reasoning](https://md.research.elamur.ai/free/en/papers/bouchekif-2026-qias-2026-overview-shared-task/). arXiv preprint arXiv:2606.13756. [↗](https://arxiv.org/abs/2606.13756)
8. Abdessalam Bouchekif, Shahd Gaben, Samer Rashwani, Somaya Eltanbouly, Mutaz Al-Khatib, Heba Sbahi, Mohammed Ghaly, and Emad Mohamed. 2026b. [Mawarith: A dataset and benchmark for legal inheritance reasoning with llms](https://md.research.elamur.ai/free/en/papers/bouchekif-2026-mawarith-dataset-benchmark-legal-inheritance/). arXiv preprint arXiv:2603.07539. [↗](https://arxiv.org/abs/2603.07539)
9. Abdessalam Bouchekif, Samer Rashwani, Emad Soliman Ali Mohamed, Mutaz Alkhatib, Heba Sbahi, Shahd Gaben, Wajdi Zaghouani, Aiman Erbad, and Mohammed Ghaly. 2025a. [QIAS 2025: Overview of the shared task on islamic inheritance reasoning and knowledge assessment](https://md.research.elamur.ai/free/en/papers/bouchekif-2025-qias-2025-overview-shared-task/). In Proceedings of The Third Arabic Natural Language Processing Conference: Shared Tasks, pages 851–860, Suzhou, China. Association for Computational Linguistics.
10. Abdessalam Bouchekif, Samer Rashwani, Heba Sbahi, Shahd Gaben, Mutaz Al Khatib, and Mohammed Ghaly. 2025b. [Assessing large language models on islamic legal reasoning: Evidence from inheritance law evaluation](https://md.research.elamur.ai/free/en/papers/bouchekif-2025-assessing-large-language-models-islamic/). In Proceedings of The Third Arabic Natural Language Processing Conference, pages 246–257, Suzhou, China. Association for Computational Linguistics.
11. Shiqi Chen, Yiran Zhao, Jinghan Zhang, I-Chun Chern, Siyang Gao, Pengfei Liu, and Junxian He. 2023. FELM: Benchmarking factuality evaluation of large language models. In Thirty-seventh Conference on Neural Information Processing Systems Datasets and Benchmarks Track. [find](https://scholar.google.com/scholar?q=Shiqi+Chen%2C+Yiran+Zhao%2C+Jinghan+Zhang%2C+I%2DChun+Chern%2C+Siyang+Gao%2C+Pengfei+Liu%2C+and+Junxian+He%2E+2023%2E+FELM%3A+Benchmarking+factuality+evaluation+of+large+language+models%2E+In+Thirty%2Dseventh+Conference+on+N)
12. David Dale, Elena Voita, Janice Lam, Prangthip Hansanti, Christophe Ropers, Elahe Kalbassi, Cynthia Gao, Loïc Barrault, and Marta R. Costa-jussà. 2023. HalOmi: A manually annotated benchmark for multilingual hallucination and omission detection in machine translation. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 638–653, Singapore. Association for Computational Linguistics. [↗](https://doi.org/10.48550/arxiv.2305.11746)
13. Passant Elchafei and Mervat Abu-Elkheir. 2025. Hallucination detectives at SemEval-2025 task 3: Span-level hallucination detection for LLM-generated answers. In Proceedings of the 19th International Workshop on Semantic Evaluation (SemEval-2025), pages 601–606, Vienna, Austria. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Passant+Elchafei+and+Mervat+Abu%2DElkheir%2E+2025%2E+Hallucination+detectives+at+SemEval%2D2025+task+3%3A+Span%2Dlevel+hallucination+detection+for+LLM%2Dgenerated+answers%2E+In+Proceedings+of+the+19th+International+W)
14. Somaya Eltanbouly, Heba Sbahi, Samer Rashwani, Abdessalam Bouchekif, Mutaz Al-Khatib, Shahd Gaben, and Mohammed Ghaly. 2026. What makes a good fiqh retriever? answer retrieval for arabic islamic jurisprudence. arXiv preprint arXiv:2608.20246. [↗](https://arxiv.org/abs/2608.20246)
15. Fanar Team, Ummar Abbas, Mohammad Shahmeer Ahmad, Firoj Alam, Enes Altinisik, Ehsannedin Asgari, Yazan Boshmaf, Sabri Boughorbel, Sanjay Chawla, Shammur Chowdhury, Fahim Dalvi, Kareem Darwish, Nadir Durrani, Mohamed Elfeky, Ahmed Elmagarmid, Mohamed Eltabakh, Masoomali Fatehkia, Anastasios Fragkopoulos, Maram Hasanain, and 23 others. 2025. [Fanar: An arabic-centric multimodal generative AI platform](https://md.research.elamur.ai/free/en/papers/abbas-2025-fanar-arabic-centric-multimodal-generative/). Preprint, arXiv:2501.13944. [↗](https://arxiv.org/abs/2501.13944)
16. Shahd Gaben, Heba Sbahi, Samer Rashwani, Abdessalam Bouchekif, Mutaz Al-Khatib, Emad Mohamed, Somaya Eltanbouly, and Mohammed Ghaly. 2026. Islamicturathbench: A multi-task, multi-discipline benchmark for evaluating large language models on the islamic scholarly tradition (turath). arXiv preprint arXiv:2608.04703. [↗](https://arxiv.org/abs/2608.04703)
17. Lei Huang, Weijiang Yu, Weitao Ma, Weihong Zhong, Zhangyin Feng, Haotian Wang, Qianglong Chen, Weihua Peng, Xiaocheng Feng, Bing Qin, and Ting Liu. 2025a. [A survey on hallucination in large language models: Principles, taxonomy, challenges, and open questions](https://md.research.elamur.ai/free/en/papers/huang-2023-survey-hallucination-large-language-models/). ACM Transactions on Information Systems, 43(2). [↗](https://doi.org/10.1145/3703155)
18. Lei Huang, Weijiang Yu, Weitao Ma, Weihong Zhong, Zhangyin Feng, Haotian Wang, Qianglong Chen, Weihua Peng, Xiaocheng Feng, Bing Qin, and 1 others. 2025b. [A survey on hallucination in large language models: Principles, taxonomy, challenges, and open questions](https://md.research.elamur.ai/free/en/papers/huang-2023-survey-hallucination-large-language-models/). ACM Transactions on Information Systems, 43(2):1– 55.
19. Ziwei Ji, Nayeon Lee, Rita Frieske, Tiezheng Yu, Dan Su, Yan Xu, Etsuko Ishii, Ye Jin Bang, Andrea Madotto, and Pascale Fung. 2023. [Survey of hallucination in natural language generation](https://md.research.elamur.ai/free/en/papers/ji-2022-survey-hallucination-natural-language-generation/). ACM Computing Surveys, 55(12):1–38. [↗](https://doi.org/10.1145/3571730)
20. Junyi Li, Xiaoxue Cheng, Xin Zhao, Jian-Yun Nie, and Ji-Rong Wen. 2023. [HaluEval: A large-scale hallucination evaluation benchmark for large language models](https://md.research.elamur.ai/free/en/papers/li-2023-halueval-large-scale-hallucination-evaluation/). In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 6449–6464, Singapore. Association for Computational Linguistics.
21. Stephanie Lin, Jacob Hilton, and Owain Evans. 2022. [TruthfulQA: Measuring how models mimic human falsehoods](https://md.research.elamur.ai/free/en/papers/lin-2022-truthfulqa-measuring-how-models-mimic/). In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 3214–3252, Dublin, Ireland. Association for Computational Linguistics.
22. Marryam Yahya Mohammed, Sama Ayman Ali, Salma Khaled Ali, Ayad Abdul Majeed, and Ensaf Hussein Mohamed. 2025. Aftina: enhancing stability and preventing hallucination in ai-based islamic fatwa generation using llms and rag. Neural Computing and Applications, 37(25):20957–20982. [↗](https://doi.org/10.1007/s00521-025-11229-y)
23. Hamdy Mubarak, Hend Al-Khalifa, and Khaloud Suliman Alkhalefah. 2024. [Halwasa: Quantify and analyze hallucinations in large language models: Arabic as a case study](https://md.research.elamur.ai/free/en/papers/mubarak-2024-halwasa-quantify-analyze-hallucinations-large/). In Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024), pages 8008–8015, Torino, Italia. ELRA and ICCL.
24. Hamdy Mubarak, Rana Malhas, Watheq Mansour, Abubakr Mohamed, Mahmoud Fawzi, Majd Hawasly, Tamer Elsayed, Kareem Mohamed Darwish, and Walid Magdy. 2025. [Islamiceval 2025: The first shared task of capturing llms hallucination in islamic content](https://md.research.elamur.ai/free/en/papers/mubarak-2025-islamiceval-2025-first-shared-task/). In Proceedings of The Third Arabic Natural Language Processing Conference: Shared Tasks, pages 480–493.
25. 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](https://md.research.elamur.ai/free/en/papers/niu-2024-ragtruth-hallucination-corpus-developing-trustworthy/). In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 10862–10878, Bangkok, Thailand. Association for Computational Linguistics. [↗](https://doi.org/10.48550/arxiv.2401.00396)
26. CH-Wang Sky, Benjamin Van Durme, Jason Eisner, and Chris Kedzie. 2024. Do androids know they’re only dreaming of electric sheep? In Findings of the Association for Computational Linguistics: ACL 2024, pages 4401–4420. [find](https://scholar.google.com/scholar?q=CH%2DWang+Sky%2C+Benjamin+Van+Durme%2C+Jason+Eisner%2C+and+Chris+Kedzie%2E+2024%2E+Do+androids+know+they%E2%80%99re+only+dreaming+of+electric+sheep%3F+In+Findings+of+the+Association+for+Computational+Linguistics%3A+ACL+2024%2C)
27. Raul Vazquez, Timothee Mickus, Elaine Zosa, Teemu Vahtola, J”org Tiedemann, Aman Sinha, Vincent Segonne, Fernando Sanchez-Vega, Alessandro Raganato, Jindřich Libovick’y, Jussi Karlgren, Shaoxiong Ji, Jindřich Helcl, Liane Guillou, Ona De Gibert, Jaione Bengoetxea, Joseph Apidianaki, and Marianna Apidianaki. 2025. SemEval-2025 task 3: Mu-SHROOM, the multilingual shared-task on hallucinations and related observable overgeneration mistakes. In Proceedings of the 19th International Workshop on Semantic Evaluation (SemEval-2025), pages 2472–2497, Vienna, Austria. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Raul+Vazquez%2C+Timothee+Mickus%2C+Elaine+Zosa%2C+Teemu+Vahtola%2C+J%E2%80%9Dorg+Tiedemann%2C+Aman+Sinha%2C+Vincent+Segonne%2C+Fernando+Sanchez%2DVega%2C+Alessandro+Raganato%2C+Jind%C5%99ich+Libovick%E2%80%99y%2C+Jussi+Karlgren%2C+Shaoxiong+Ji%2C+)
28. Hanzhi Zhang, Sumera Anjum, Heng Fan, Weijian Zheng, Yan Huang, and Yunhe Feng. 2025. Poly-FEVER: A multilingual fact verification benchmark for hallucination detection in large language models. arXiv preprint arXiv:2503.16541. A Hallucination in Large Language Models and Related Resources Definition. LLMs can generate fluent and convincing answers that are not necessarily factually [↗](https://arxiv.org/abs/2503.16541)

*Updated 1 November 2026*