Large Language Models (LLMs) have rapidly transitioned from experimental research artifacts to widely deployed socio-technical systems used in journalism, law, education, healthcare, and scientific research. Alongside their demonstrated fluency and versatility, these models exhibit a persistent and consequential limitation: hallucination, broadly understood as the generation of content that is syntactically plausible but factually incorrect, unverifiable, or internally inconsistent. High-profile incidents—including fabricated legal precedents, false accusations against real individuals, and erroneous scientific claims—have underscored the real-world risks associated with unmitigated hallucinations. This article provides a comprehensive, system-level examination of hallucinations in LLMs, synthesizing empirical case studies, theoretical taxonomies, detection methodologies, mitigation strategies, and evaluation frameworks. Drawing on recent academic literature and documented failures in deployed systems, the paper categorizes hallucinations into factual, logical, contextual, and self-contradictory forms, and analyzes their underlying mechanisms in model training, prompting, and deployment contexts. Particular emphasis is placed on black-box and zero-resource detection approaches, including SelfCheckGPT, Chain-of-Thought consistency methods, and alignment-based evaluation metrics such as GPTScore, GEval, and AlignScore. The discussion critically examines the limitations of current benchmarks and the epistemic challenges of evaluating truthfulness in generative systems. The article concludes by outlining open research challenges and ethical considerations, arguing that hallucination mitigation should be framed not as a problem of complete elimination, but as a continuous risk-management process integrating technical, institutional, and human-in-the-loop safeguards.

Large Language Models, AI hallucination, factual consistency, hallucination detection, evaluation metrics, AI reliability

1. S. K. Bordoloi, “The hilarious & horrifying hallucinations of AI,” 2 July 2023. Available: https://www.sify.com/ai-analytics/the-hilarious-andhorrifying-hallucinations-of-ai/.

2. J. Brodkin, “Lawyer cited 6 fake cases made up by ChatGPT; judge calls it ‘unprecedented’,” 31 May 2023. Available: https://arstechnica.com/tech-policy/2023/05/lawyer-cited-6-fake-cases-made-up-by-chatgpt-judge-calls-it-unprecedented/.

3. P. Prakash, “ChatGPT falsely accused a mayor of bribery when he was actually the whistleblower,” 6 April 2023. Available: https://fortune.com/2023/04/05/chatgpt-falsely-accused-australian-mayor-bribery-openai-defamation/.

4. S. Cole, “‘Life or Death:’ AI-Generated Mushroom Foraging Books Are All Over Amazon,” 29 August 2023. Available: https://www.404media.co/ai-generated-mushroom-foraging-books-amazon/.

5. J. Vincent, “Google’s AI chatbot Bard makes factual error in first demo,” 8 February 2023. Available: https://www.theverge.com/2023/2/8/23590864/google-ai-chatbot-bard-mistake-error-exoplanet-demo.

6. OpenAI, “GPT-4,” OpenAI, 2023.

7. F. Wang, “A Survey of Multimodal LLM Hallucination,” January 2024. Available: https://github.com/ZurichRain/AGIHallucination/.

8. P. Manakul, A. Liusie, and M. J. F. Gales, “SelfCheckGPT: Zero-Resource Black-Box Hallucination Detection for Generative Large Language Models,” Cornell University, 2023.

9. N. Mündler, J. He, S. Jenko, and M. Vechev, “Self-contradictory Hallucinations of Large Language Models: Evaluation, Detection and Mitigation,” Cornell University, 2024.

10. J. Fu, S.-K. Ng, Z. Jiang, and P. Liu, “GPTScore: Evaluate as You Desire,” Cornell University, 2023.

11. Y. Liu et al., “GEval: NLG Evaluation using GPT-4 with Better Human Alignment,” Cornell University, 2023.

12. R. Friel and A. Sanyal, “ChainPoll: A High Efficacy Method for LLM Hallucination Detection,” Cornell University, 2023.

13. Z. Ji et al., “RHO: Reducing Hallucination in Open-Domain Dialogues with Knowledge Grounding,” 2023.

14. N. Dziri et al., “Neural Path Hunter: Reducing Hallucination in Dialogue Systems via Path Grounding,” EMNLP, 2021.

15. K. Shuster et al., “Retrieval Augmentation Reduces Hallucination in Conversation,” 2021.

16. H. Rashkin et al., “Increasing Faithfulness in Knowledge-Grounded Dialogue with Controllable Features,” ACL-IJCNLP, 2021.

17. W. Sun et al., “Contrastive Learning Reduces Hallucination in Conversations,” AAAI, 2023.

18. Y. Zha et al., “AlignScore: Evaluating Factual Consistency with a Unified Alignment Function,” ACL, 2023.

19. L. Reynolds and K. McDonell, “Prompt Programming for Large Language Models: Beyond the Few-Shot Paradigm,” Cornell University, 2021.