Persistent Anti-Muslim Bias in Large Language Models
Open accessThe DOI matches its Crossref record; 13 of the 13 references it cites were found in a registry or an index.
Abstract
It has been observed that large-scale language models capture undesirable societal biases, e.g. relating to race and gender; yet religious bias has been relatively unexplored. We demonstrate that GPT-3, a state-of-the-art contextual language model, captures persistent Muslim-violence bias. We probe GPT-3 in various ways, including prompt completion, analogical reasoning, and story generation, to understand this anti-Muslim bias, demonstrating that it appears consistently and creatively in different uses of the model and that it is severe even compared to biases about other religious groups. For instance, Muslim is analogized to terrorist in 23% of test cases, while Jewish is mapped to its most common stereotype, money, in 5% of test cases. We quantify the positive distraction needed to overcome this bias with adversarial text prompts, and find that use of the most positive 6 adjectives reduces violent completions for Muslims from 66% to 20%, but which is still higher than for other religious groups.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
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
“Pre-trained on massive amounts of text on the Internet, LMs are known to produce linguistically fawed, factually incorrect, or even ethically problematic text at times [1,24].”
From CoAuthor: Designing a Human-AI Collaborative Writing Dataset for Exploring Language Model Capabilities (2022) · cites this paper as [1]
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
“LLMs and other AI tools may have a number of biases that could arise from various factors, such as unbalanced training datasets74–77.”
From On opportunities and challenges of large multimodal foundation models in education (2025) · cites this paper as [75]
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
“it is also a liability; all AI systems might inherit the same problematic biases of a few foundation models [Bolukbasi et al. 2016; Caliskan et al. 2017; Abid et al. 2021, inter alia]”
From On the Opportunities and Risks of Foundation Models (2021) · cites this paper as Abid et al. 2021
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“Previous work has shown that language models tend to exhibit human-like biases with respect to specific protected groups such as gender, ethnicity or religious orientation3–6.”
From Generative language models exhibit social identity biases (2024) · cites this paper as [4]
In the citing paper
“If a medical LLM has been trained on data that contains misinformation or biased content, LLMs deployed for clinical decision-making are at risk of reproducing these problematic contents [120,124–126].”
From Bias in medical AI: Implications for clinical decision-making (2024) · cites this paper as [126]
1 more claim 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.
The self-check needs the paper’s full text.
Citations
Cites
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., & Polosukhin, I. (2017) · arXiv · doi:10.48550/arxiv.1706.03762 · Cited by 120,404 · cited 1×
“These languagemodels have evolved, from learning individual word vectors with single-layer models [8], to more complex language generation architectures such as recurrent neural networks [5] and most recently transformers [4, 6, 14].”
BackgroundEfficient Estimation of Word Representations in Vector Space
Mikolov, T., Chen, K., Corrado, G., & Dean, J. (2013) · arXiv · doi:10.48550/arxiv.1301.3781 · Cited by 26,568 · cited 1×
“These languagemodels have evolved, from learning individual word vectors with single-layer models [8], to more complex language generation architectures such as recurrent neural networks [5] and most recently transformers [4, 6, 14].”
BackgroundLanguage 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 · cited 4×
“These languagemodels have evolved, from learning individual word vectors with single-layer models [8], to more complex language generation architectures such as recurrent neural networks [5] and most recently transformers [4, 6, 14].”
BackgroundMethodInfluentialSelf-Critical Sequence Training for Image Captioning
Rennie, S. J., Marcheret, E., Mroueh, Y., Ross, J., & Goel, V. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.131 · Cited by 2,112 · cited 1×
“These descriptions are generated by passing images through a pretrained image captioning model [11], and fine-tuning GPT-3 to transform the short captions into long descriptions, as shown in Fig.”
MethodReformer: The Efficient Transformer
Kitaev, N., Kaiser, Ł., & Levskaya, A. (2020) · arXiv · doi:10.48550/arxiv.2001.04451 · Cited by 1,435 · cited 1×
“These languagemodels have evolved, from learning individual word vectors with single-layer models [8], to more complex language generation architectures such as recurrent neural networks [5] and most recently transformers [4, 6, 14].”
BackgroundSemi-supervised Sequence Learning
Dai, A. M., & Le, Q. V. (2015) · arXiv · doi:10.48550/arxiv.1511.01432 · Cited by 994
StereoSet: Measuring stereotypical bias in pretrained language models
Nadeem, M., Bethke, A., & Reddy, S. (2020) · arXiv · doi:10.48550/arxiv.2004.09456 · Cited by 643 · cited 2×
“As a result, even though the various language models have different architectures, since they are trained on similar corpora of text, they often learn similar biases [9].”
BackgroundUniversal Adversarial Triggers for Attacking and Analyzing NLP
Wallace, E., Feng, S., Kandpal, N., Gardner, M., & Singh, S. (2019) · Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP) · doi:10.18653/v1/d19-1221 · Cited by 642
Identifying and Reducing Gender Bias in Word-Level Language Models
Bordia, S., & Bowman, S. R. (2019) · Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Student Research Workshop · doi:10.18653/v1/n19-3002 · Cited by 219 · cited 1×
“Previous work has explored the biases that are ultimately learned by language models, with the primary focus being racial and gender stereotypes [2, 3, 7, 12].”
BackgroundThe Woman Worked as a Babysitter: On Biases in Language GenerationPreprint
Sheng, E., Chang, K.-W., Natarajan, P., & Peng, N. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1909.01326 · Cited by 205 · cited 2×
“In [12], the authors made similar observations about language generation models.”
Background
Research works printed but not linked: 6
- 7. T. Bolukbasi, K.-W. Chang, J. Y. Zou, V. Saligrama, A. T. Kalai, Advances in neural information processing systems 29 , 4349 (2016).
- 9. S. Bordia, S. R. Bowman, arXiv preprint arXiv:1904.03035 (2019). · arXiv
- 2. A. M. Dai, Q. V. Le, Advances in neural information processing systems 28 , 3079 (2015).
- 10. K. Lu, P. Mardziel, F. Wu, P. Amancharla, A. Datta, Logic, Language, and Security (Springer, 2020), pp. 189-202.
- 3. A. Vaswani, et al. , Advances in neural information processing systems (2017), pp. 59986008.
- 14. E. Wallace, S. Feng, N. Kandpal, M. Gardner, S. Singh, arXiv preprint arXiv:1908.07125 (2019). · arXiv
Cited by
Items 801–900 of 986
StylisticBias: A Few Human Visual Cues Drive Most Social Biases in MLLMsPreprint
Kolli, S., Cavelius, T., Nikeghbal, N., Dalal, S., & Diesner, J. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2606.20527 · Cited by 1
“…been extensively documented in large language models, which matters as biases reproduce and amplify societal stereotypes embedded in text corpora (Shrawgi et al., 2024; Ostrow and Lopez, 2025; Sheng et al., 2019; Abid et al., 2021; Parrish et al., 2022; You et al., 2026; Nikegh-bal et al., 2025).”
Khan, M. S., Umer, H., Mahmud, H., & Rothenberg, S. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7220762 · Cited by 1
BavGround: A Benchmark for Regional Cultural Grounding and Dialect Competence in BavarianPreprint
John, J., Hoffmann, M., Fillies, J., Hedderich, M. A., & Plank, B. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2608.12894 · Cited by 1
“For example, Naous et al. (2024) study culturally situated continuations in religious and social contexts, while broader work examines how language models reproduce culturally dependent stereotypes (Abid et al., 2021; Kirk et al., 2021; Kumar et al., 2025).”
Veracity Labels Flip the Emotional Tone of Language Models but Leave Racial Disparities IntactPreprint
AlDahoul, N., Tan, M. J., & Zaki, Y. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7584919 · Cited by 1
IslamicMMLU: A Benchmark for Evaluating LLMs on Islamic Knowledge
Abdelaal, A., Haffar, M. N. A., Fawzi, M., & Magdy, W. (2026) · arXiv · doi:10.48550/arxiv.2603.23750
“Abid et al. (2021) provide foundational evidence of persistent anti-Muslim bias in GPT-3.”
On Epistemic Diversity in Large Language Models
Kirsten, E., Krämer, N., & Zafar, M. B. (2026) · arXiv · doi:10.48550/arxiv.2609.04835
“The representational lens asks whether all groups are given proportional representation without misrepresentations such as negative stereotypes (Schw¨obel et al., 2023; Mehrabi et al., 2021; Abid et al., 2021; Bolukbasi et al., 2016; Guo & Caliskan, 2021).”
Uncovering Gaps in How Humans and LLMs Interpret Subjective Language
Jones, E., Patrawala, A., & Steinhardt, J. (2025) · arXiv · doi:10.48550/arxiv.2503.04113
“…more scalable human supervision. from misinterpreting human instructions; LLMs can propagate stereotypes (Sheng et al., 2019; Blodgett et al., 2021; Abid et al., 2021), hallucinate (Ji et al., 2023; Min et al., 2023), and overreact to unimportant parts of instructions (Jones & Steinhardt, 2022;…”
Chat-H I : Exploring Indian National Identity Through Machine-Generated Text
Martelli, J., & Parekh, S. (2021) · Leonardo · doi:10.1162/leon_a_02141
EXIST 2021: Inducing Bias in Deep Learning Models.
Garrido-Muñoz, I., Montejo‐Ráez, A., & Santiago, F. J. M. (2021) · IberLEF@SEPLN
ДИННИН, Д. Д., & БУБИНГЕР, Х. (2021) · Международный форум по информации · doi:10.36535/0203-6460-2021-03-2
Semantic Search over Wikipedia Documents Meaning of Queries Based Pre-Trained Language Model
Tharun, P., & TJPRC (2021) · International journal of computer science engineering and information technology research · doi:10.24247/ijcseitrdec20212
DialCL-Bias: A Semantic and Contextual Framework to Identify Social Bias in Dialogue
Cai, Z., Wu, D., & Li, P. (2023) · Conference proceedings/Conference proceedings - IEEE International Conference on Systems, Man, and Cybernetics · doi:10.1109/smc53992.2023.10394325
“Studies have shown that language models exhibit bias regarding gender [5], race [6], and occupation.”
BackgroundBeyond the Hype: A Critical Evaluation of ChatGPT's Capabilities for Mathematical Calculations
Quansah, E. M. (2024) · International Conference on Artificial Intelligence and Big Data (AIBD) · doi:10.5121/csit.2024.140407
“ChatGPT and similar large language models are referred to as generative AI[9], [13].”
BackgroundShaping the future: responsible regulation of generative artificial intelligence
Buzu, I. (2024) · Vector European · doi:10.52507/2345-1106.2024-1.03
FairBelief - Assessing Harmful Beliefs in Language Models
Setzu, M., Manerba, M. M., Minervini, P., & Nozza, D. (2024) · Workshop on Trustworthy Natural Language Processing (TrustNLP) · doi:10.18653/v1/2024.trustnlp-1.3
“Moreover, recent studies have documented the anti-Muslim sentiment exhibited by GPT-3 (Abid et al., 2021), which generated toxic and abusive text when interrogated with prompts containing references to Islam and Muslims.”
BackgroundCharacterizing Stereotypical Bias from Privacy-preserving Pre-Training
Arnold, S., Gröbner, R., & Schreiner, A. (2024) · Workshop on Privacy in Natural Language Processing (PrivateNLP) · doi:10.18653/v1/2024.privatenlp-1.3
Tanaka, K., Sasano, R., & Takeda, K. (2024) · arXiv (Cornell University) · doi:10.48550/arxiv.2408.04293
“For example, language models are known to learn a broad spectrum of biases, including those related to gender (Kirk et al., 2021; Lucy and Bamman, 2021), nationality (Narayanan Venkit et al., 2023), religion (Abid et al., 2021), and race (Field et al., 2021).”
BackgroundThe Metaverse and Legal Status of Intelligent Agents
Buzu, I. (2024) · Studii Juridice Universitare · doi:10.54481/sju.2023.2.11
Augmented Reality Research: Benefit or Detriment for Neurodiverse People
Jones, D., Gračanin, D., & Azab, M. (2024) · IEEE International Symposium on Mixed and Augmented Reality Adjunct (ISMAR-Adjunct) · doi:10.1109/ismar-adjunct64951.2024.00015
“Research has shown that most biases occur during model generation training and fine-tuning stages [1,8,10,11,14].”
Intersectional Bias in Causal Language Models
Magee, L., Ghahremanlou, L., Soldatić, K., & Robertson, S. (2024) · doi:10.32920/27956337.v1
Developing an occupational prestige scale using Large Language ModelsPreprint
de Vries, R., Hill, M. J., & Ruis, L. (2025) · SocArXiv (OSF Preprints) · doi:10.31235/osf.io/ct4vz_v2
AI and NLP-Based Question Generation Using the T5 Model for Education
Balamurugan, A., Bharathidasan, S., Arputha Joshwa, B., Rathinapriya, V., & Balaji, V. (2025) · International Research Journal on Advanced Engineering and Management (IRJAEM) · doi:10.47392/irjaem.2025.0251
Stockwell, G., & Wang, Y. (2025) · Cambridge University Press eBooks · doi:10.1017/9781009294850.022
Gender-Neutral Large Language Models for Medical Applications: Reducing Bias in PubMed Abstracts
Schaefer, E., & Roberts, K. E. (2025) · Workshop on Biomedical Language Processing (BioNLP) · doi:10.18653/v1/2025.bionlp-1.11
How Generative AI Works and How It Fails
Narayanan, A., & Kapoor, S. (2025) · doi:10.21428/2c646de5.70cfbc85
de Sena, R. L. R., Souza, M., Santana, A. S., & de Melo, J. L. L. (2025) · Anais do Workshop sobre as Implicações da Computação na Sociedade (WICS) · doi:10.5753/wics.2025.9056
Is AI coming for your creative job? Maybe not – with some human intervention
Soudi, A., Adamson, G., Escandón, L., & Asaleh, R. (2025) · The Conversation · doi:10.64628/aam.ccg3dc6ty
(2025) · Medien & Kommunikationswissenschaft · doi:10.5771/1615-634x-2025-3-472
Medien & Kommunikationswissenschaft: M & K
Weiß, A.-N., Lünenborg, M., Weiß, A.-N., Bayramoğlu, Y., & Uth, B. (2025) · Medien & Kommunikationswissenschaft · doi:10.5771/1615-634x-2025-3
Pekerjaan kreatif bakal digantikan AI? Kepiawaian penggunanya tetap jadi kunci
Soudi, A., Adamson, G., Escandón, L., & Asaleh, R. (2025) · The Conversation · doi:10.64628/aan.3s6xgsrhq
Harnessing ChatGPT for Innovation and Creativity: Navigating Ethical Dilemmas and Governance Gaps
Pahuja, A., Kaur, S., & Hassan, R. (2025) · IGI Global eBooks · doi:10.4018/979-8-3373-0877-7.ch011
Unmaking AI: Engaging Critically and Creatively with Generative AI
Munn, L., Khan, A. H., Hettiachchi, D., Sabie, S., Ghahremanlou, L., Saghir, S., Lambourne, N., & Magee, L. (2024) · Australian Conference on Human-Computer Interaction (OzCHI) · doi:10.1145/3726986.3727943
“Numerous studies have highlighted bias across gender, race, class, disability and other categories [2, 6, 9], which translate into social harms as models are adopted and employed.”
Ruminative Thought Content Changes by Year: A Computational Linguistic Analysis
Strege, M. V., Krawczak, R. A., Kim, Y., Semerod, J., Brush, L., & Siegle, G. J. (2025) · Cognitive Therapy and Research · doi:10.1007/s10608-025-10666-2
Developing A Framework to Support Human Evaluation of Bias in Generated Free Response TextPreprint
Healey, J., Byrum, L., Akhtar, M. N., Bhargava, S., & Sinha, M. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2505.03053
Perspective Dial: Measuring Perspective of Text and Guiding LLM OutputsPreprint
Kim, T., Mau, S.-C., & Vesey, K. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2506.23377
“Biases and associated perspectives manifest as gender, racial, cultural, and socio-economic stereotypes (Guo et al., 2024; Abid et al., 2021).”
Addressing Stereotypes in Large Language Models: A Critical Examination and MitigationPreprint
Kazi, F. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2511.21711
Understanding Down Syndrome Stereotypes in LLM-Based Personas
Wu, C., Pan, M., Wang, P., Nibras, N., Li, M., Yuan, D., Wang, Z., He, J., Ali, M. A. S., & Prpa, M. (2026) · Designing Interactive Systems Conference · doi:10.1145/3800645.3812894
“This, if not mediated, can potentially lead to oversimplified content and biased behaviour towards certain social groups based off of disability [70], gender [5, 52, 75, 94, 98], occupation [98], race [5, 94], and religion [2, 48].”
Language Models That Walk the Talk: A Framework for Formal Fairness CertificatesPreprint
Chen, D. J., Ladner, T., Mhadhbi, A. R., & Althoff, M. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2505.12767
“…in both the pre-training and language model training phase leads to a drop in model accuracy, indicating that such features can play a legitimate role in prediction, highlighting the necessity of our proposed fairness score to assess and manage potential bias rather than outright exclusion.”
FOUNDATIONS OF TRUSTWORTHY DEEP LEARNING: FAIRNESS, ROBUSTNESS, AND EXPLAINABILITY
Nanda, V. (2024) · University Libraries (University of Maryland) · doi:10.13016/twct-cnxl
Gender-Neutral Large Language Models for Medical Applications: Reducing Bias in PubMed AbstractsPreprint
Schaefer, E., & Roberts, K. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2501.06365
From Charts to Fair Narratives: Uncovering and Mitigating Geo-Economic Biases in Chart-to-TextPreprint
Mahbub, R., Islam, M. S., Nayeem, M. T., Laskar, M. T. R., Rahman, M., Joty, S., & Hoque, E. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2508.09450
“To mitigate geo-economic bias in VLM responses, we adopted an inference-time prompt-based approach inspired by Abid et al. (2021); Narayanan Venkit et al. (2023), which utilizes positive distractions.”
New approaches to studying the cognitive impact of a CAI tool on Chinese interpreting trainees
Du, Z., & Ricardo, M. M. (2024) · Archivio istituzionale della ricerca (Alma Mater Studiorum Università di Bologna)
Investigation of the features of using large language models for tabular data
Шубаков, В. С., & Shubakov, V. S. (2024) · Electronic scientific archive of UrFU (Ural Federal University)
KoGPT-2 생성 문장의 주거 관련 편견 분석 및 감소 방법 연구
서나래 (2022) · Seoul National University Open Repository (Seoul National University)
Woman Is To Nurse As Man Is To... Nurse Too? : Uncovering Structural Bias in GPT-2
Jiayu, S. (2025) · DiVA (Digitala Vetenskapliga Arkivet)
Mayer, C. (2024) · MADOC (University of Mannheim)
Lippens, L. (2023) · VUBIR (Vrije Universiteit Brussel)
de Weerd, D. (. (2023) · Research Repository (Delft University of Technology)
TagDebias: Entity and Concept Typing for Social Bias Mitigation in Pretrained Language Models
Moslemi, M. (2024) · PolyPublie (École Polytechnique de Montréal)
Essays on Law and Natural Language Processing
Akdemir, E. I. (2023) · eScholarship (California Digital Library)
QueerBench: Quantifying Discrimination in Language Models towards Queer Identities
Sosto, M. (2023) · AMS Laurea Institutional Research Repository (University of Bologna)
Pisciotta, G. (2021) · Electronic Theses and Dissertations Repository (University of Pisa)
Timmerman, K. (. (2023) · Research Repository (Delft University of Technology)
Analysing social biases toward migrant groups encoded in language models
Sorato, D. (2024) · Repositori digital de la UPF (Universitat Pompeu Fabra)
Data-Centric Methods for Decentralizing Large Language Models
Gururangan, S. (2024) · ResearchWorks at the University of Washington (University of Washington)
Measuring and Manipulating State Representations in Neural Language Models
Li, B. Z. (2023) · DSpace@MIT (Massachusetts Institute of Technology)
Teknologiapedagogiikkaa kielten opetukseen
(2025) · Jyväskylä University Digital Archive (University of Jyväskylä)
Salazar, G. P. (2025) · Dipòsit Digital de Documents de la UAB (Universitat Autònoma de Barcelona)
More Robust Learning Guarantees via Active Dataset Construction
Rittler, N. (2025) · eScholarship (California Digital Library)
김병필 (2023) · Seoul National University Open Repository (Seoul National University)
Technical methods for governing AI agents
Chan, A. (2025) · doi:10.71781/10081
Biased by design: Leveraging Inherent AI Biases to Enhance Critical Thinking of News Readers
Zavolokina, L., Sprenkamp, K., Katashinskaya, Z., & Jones, D. G. (2025) · AIS Electronic Library (AISeL)
Who Gets to Be Biased? How AI Supply Chains Overrepresent the West and Marginalize the Rest
Puutio, A., & Lin, P. K. (2025) · Oxford University Press eBooks · doi:10.1093/9780198945215.003.0182
Pinheiro, J. H. S. (2025) · Aurum Revista Multidisciplinar · doi:10.63330/armv1n5-015
Ottaviani, M., & Stahlschmidt, S. (2026) · Scientometrics · doi:10.1007/s11192-026-05857-x
Opuszko, M., & Böhm, P. (2026) · Communications in computer and information science · doi:10.1007/978-3-032-12313-8_7
Prompt-Counterfactual Explanations for Generative AI System BehaviorPreprint
Goethals, S., Provost, F. J., & Sedoc, J. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2601.03156
da Silva, M. F. (2024) · LA Referencia (Red Federada de Repositorios Institucionales de Publicaciones Científicas)
README: A Bookish History of Computing from Electronic Brains to Everything Machines
McCray, W. P. (2025) · The MIT Press eBooks · doi:10.7551/mitpress/14213.001.0001
(2025) · The MIT Press eBooks · doi:10.7551/mitpress/15301.003.0010
Myth, reality, or in between: Unveiling potential geographical biases of ChatGPT
Wijewickrema, M. (2026) · Information Research · doi:10.47989/ir31146885
“Additionally, Gross (2023) and Kaplan et al. (2024) examined gender bias, whereas Abid et al. (2021) investigated religious bias.”
AI-collaborative method for qualitative sensemaking: an application to entrepreneurial identity
Valliere, D. (2026) · International Journal of Entrepreneurial Behaviour & Research · doi:10.1108/ijebr-06-2025-0890
PRISM: Principled Reasoning for Identifying and Suppressing Model Biases at Scale
Zhu, X., Li, W., & Lin, J. (2026) · Lecture notes in computer science · doi:10.1007/978-981-95-7081-2_3
صويلح, س. (2025) · مجلة جامعة فلسطين التقنية للأبحاث · doi:10.53671/pturj.v13i04.743
Toward a Philosophical Taxonomy of Bias in Large Language Models: A SurveyPreprint
Jiang, K., Jacobs, J., Aiyar, T., Tong, T., Lin, Y., Abukhadier, Z., Khawaja, R., Sajeev, A., Kundam, A., Li, W., et al. (2026) · doi:10.36227/techrxiv.177273600.04458982/v1
Cognitive Error Correction Prompting for Effective Social Bias Mitigation in Large Language Models
Lee, J., Shin, H., & Shin, K.-S. (2025) · IEEE International Conference on Big Data (BigData) · doi:10.1109/bigdata66926.2025.11402427
“However, because these models learn directly from human-written data, they inevitably absorb and reproduce societal biases, stereotypes, and ideological distortions embedded in the texts [1] [8] [34] [55].”
Date: dynamic attention-based toxicity elimination mechanism for safe image generation
Cha, Y., Park, J., & Lee, Y. (2026) · Multimedia Systems · doi:10.1007/s00530-025-02166-w
Bouhlaoui, A. (2025) · International Journal of Current Research in Science Engineering & Technology · doi:10.30967/ijcrset/aadil-bouhlaoui/205
Emergent Biases in Large Language Models: A Critical Review of Taxonomy and Evaluations
Ismail, S., & Aygün, R. S. (2025) · IEEE International Conference on Cognitive Machine Intelligence (CogMI) · doi:10.1109/cogmi67134.2025.00036
“Similarly, Abid et al. [19] showed that GPT-3 associated Muslims with terrorism at much higher rates than smaller predecessors.”
InfluentialCreative Agency and Ethical Boundaries: A Documentary Analysis of AI in the Creative Industries
Zengo, L. (2026) · Interdisciplinary Academic and Research Journal · doi:10.60027/iarj.2026.e290412
“Abid et al. (2021) extended this analysis by demonstrating that GPT3 is 23 percent more likely to associate Islam with violence compared to other religions, mirroring entrenched media stereotypes.”
SectEval: Evaluating the Latent Sectarian Preferences of Large Language ModelsPreprint
Maheshwari, A., Gajkeshwar, A., -, K. S., & Patel, V. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2603.12768
“Significant research has been conducted to demonstrate the prevalence of religious bias in LLMs. Prior investigations, including (Abid et al., 2021a), have documented instances of anti-Muslim bias within GPT-3, frequently manifesting in associations with violence, even when presented with neutral…”
Nurhidayati, H., Nurchayati, Z., & Sugiarto (2026) · Research Square · doi:10.21203/rs.3.rs-8994226/v1
Zhang, L., Zhao, N., & Wan, J. (2026) · Research Square · doi:10.21203/rs.3.rs-8966332/v1
Neural Story Generation via Graph-based Content Planning
陳, 宏. (2022) · Institutional Repositories DataBase (IRDB) · doi:10.15083/0002011075
Patel, N. S., & Iandoli, L. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-15501-6_13
Improving Representations for Language Modeling
Godey, N. (2024) · theses.fr (ABES) · doi:10.70675/b6a15d0ez15bez43e5za967z16a069e0a8bb
Surprising Personas: Persona Sets with Controllably Unexpected Features
Chance, C. A., & Healey, J. (2026) · CHI Conference Extended Abstracts on Human Factors in Computing Systems (CHI EA) · doi:10.1145/3772363.3798969
Building Bridges: A Participatory Framework for AI Accountability in South Asia
Islam, F., Mayeesha, T. T., & Ahmed, N. (2026) · CHI Conference Extended Abstracts on Human Factors in Computing Systems (CHI EA) · doi:10.1145/3772363.3799093
“Algorithmic fairness research has been criticized for being too Global North focused [4].”
LLM Nepotism in Organizational GovernancePreprint
Mao, S., Guo, W., Zhang, D., Zhang, C., & Cai, W. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2604.09620
“These biases can manifest as derogatory language (Beukeboom and Burgers, 2019), toxicity (Dixon et al., 2018), stereotyping (Liu et al., 2024; Abid et al., 2021; Blodgett et al., 2021), misrepresentation (Smith et al., 2022), exclusionary norms (Bender et al., 2021), and discrimination (Ferrara,…”
UMELÁ INTELIGENCIA V HISTORICKEJ VEDE NA SLOVENSKU
Drábik, J. (2026) · Historický časopis · doi:10.31577/histcaso.2026.74.1.6
Housley, W., & Dahl, P. (2026) · Emerald Group Publishing Limited eBooks · doi:10.1108/978-1-83753-878-220261011
Measuring the Machine: Evaluating Generative AI as Pluralist Sociotechical SystemsPreprint
Johnson, R. L. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2604.20545 · Cited by 1
Bias and Hallucination Evaluation in LLMs
Sathiyaseelan, R., Reshma, A. B., & Ganga, P. (2026) · Journal of Technology Informatics and Engineering · doi:10.51903/jtie.v5i1.476
From Body to Mind: Analyzing Gender Representation in Spanish Generative Language Models
Garrido-Muñoz, I., Martínez-Santiago, F. J., & Montejo‐Ráez, A. (2026) · Language Resources and Evaluation Conference (LREC) · doi:10.63317/5fz5bb6tihu3
“For instance, documented associations include stereotyping women in domestic roles and men in professional ones, or linking specific religions with violent extremism (Abid et al., 2021).”
Evaluating and Enhancing Large Language Models for Islamic Question Answering
Haque, M. M., Fahim, M. H., Selim, M. R., Haque, M. M., Amin, M. F. A., Ataullha, M., & Rahman, M. (2025) · International Conference on Computer and Information Technology (ICCIT) · doi:10.1109/iccit68739.2025.11491412
Mitigating Security Risks in Large Language Models: A Full Lifecycle Perspective
Wang, Y. K., Bai, Z., Huo, J., Wang, B., Liu, J., Cao, H., Meng, F., Yang, X., & Gao, Y. (2026) · Machine Learning · doi:10.1007/s10994-025-06930-x
“In con - trast, LLM-generated prompts (Perez et al., 2022; Yu et al., 2023 ) improved efficiency but often resulted in low-quality cases due to inherent model biases (Abid et al., 2021; Gallegos et al., 2024).”
MacArthur, M. J. (2026) · The WAC Clearinghouse; University Press of Colorado eBooks · doi:10.37514/per-b.2026.2777.2.25
Söilen, K. S. (2026) · Springer eBooks · doi:10.1007/978-3-032-23069-0_7
Murdaugh, K. M., & Kwok, I. B. (2026) · Journal of Palliative Medicine · doi:10.1177/10966218261452780
Metadata
- Authors
- Abid, A., Farooqi, M., & Zou, J.
- Affiliations
- Stanford University (US)
- McMaster University (CA)
- Venue
- Proceedings of the 2021 AAAI/ACM Conference on AI, Ethics, and Society
- Publisher
- ACM
- Published
- 2021-07-21
- Pages
- 298-306
- ISSN
- 3065-8365 (ISSN-L)
- OpenAlex
- W3184144760
- Cited by
- 476
- References
- 3
- Citations by year
- 2026: 81 · 2025: 124 · 2024: 105 · 2023: 97 · 2022: 36 · 2021: 33
- FWCI
- 61.36
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- Proceedings of the AAAI/ACM Conference on AI Ethics and Society (publishedVersion)
- arXiv (Cornell University) (submittedVersion, cc-by)
- arXiv (Cornell University) (cc-by)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Physical Sciences › Computer Science › Artificial Intelligence › Hate Speech and Cyberbullying Detection
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Keywords
- Islamophobia, religious bias, large language models, GPT-3, analogical reasoning, adversarial prompts, story generation, bias mitigation
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.
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.