Ethical and regulatory challenges of Generative AI in education: a systematic review
Open accessThe DOI matches its Crossref record; 49 of the 50 references it cites were found in a registry or an index; 21 of the 84 sources its claims cite were checked for support.
Abstract
Introduction Generative Artificial Intelligence (GenAI) is transforming education by enabling personalized learning and more efficient teaching practices. However, it raises critical ethical concerns, including data privacy, algorithmic bias, and educational inequality, requiring comprehensive regulatory frameworks and pedagogical strategies. Methods A Systematic Literature Review (SLR) was conducted, analyzing 53 peer-reviewed articles published between 2020 and 2024. The search was performed in Scopus and Web of Science using defined inclusion criteria focused on GenAI applications in education. Data were synthesized thematically and supported by theoretical frameworks from ethics, regulation, and learning sciences. Results The findings reveal that while GenAI enhances personalized feedback, instructional automation, and learning accessibility, it simultaneously introduces risks such as loss of cognitive autonomy, institutional misuse of student data, and lack of regulatory oversight. Case studies from Stanford and the University of Toronto illustrate both opportunities and limitations of GenAI adoption in higher education. Discussion GenAI can benefit education if implemented within ethical, legal, and pedagogical boundaries. The study highlights the urgency of designing inclusive regulatory frameworks, strengthening digital literacy, and integrating GenAI tools with constructivist and self-determined learning models. This review offers practical recommendations for educators, policymakers, and technologists aiming to use GenAI responsibly in educational environments.
Overview
CoChat Summary
TL;DR
This systematic review of 53 peer-reviewed articles (2020–2024) synthesizes ethical, regulatory, and educational-quality challenges of generative AI in education. It finds GenAI can enhance personalization, feedback, and accessibility, but risks cognitive dependence, data misuse, bias, and regulatory gaps, arguing for adaptive governance, digital literacy, and constructivist/SDT-aligned integration.
Key findings
- Across 53 articles (44 from Scopus, 9 from WoS), the most frequent ethical concerns were data privacy, algorithmic bias, misinformation, loss of creativity, and technological dependence; GenAI also raised academic integrity/plagiarism and student–teacher relationship concerns.
- Reported benefits include personalized feedback, instructional automation, learning accessibility, adaptive learning, and automated feedback; reported risks include a widened digital divide if access is inequitable, decreased autonomous learning/critical skills, and institutional misuse of student data.
- Regulatory analysis discusses GDPR, the EU AI Act, White House AI guidance, and Latin American national AI strategies; it reports a lack of specific regulation for AI in education in Latin America and calls for adaptive, collaborative, internationally harmonized governance.
- Case studies: Stanford used GenAI for personalized tutoring/critical thinking but highlighted digital literacy and content-verification needs; the University of Toronto allowed GenAI only as assessment support, not replacing human judgment, with auditing.
Methods
- Systematic Literature Review following Kitchenham phases and PRISMA, with three research questions on curricula/job readiness, educator/researcher empowerment, and policy adaptability/equity.
- Searches in Scopus and Web of Science using keywords “generative artificial intelligence” and “ethics,” period 2020–2024, document type article, extraction on December 4, 2024; Scopus yielded 95 records and WoS 24, six duplicates were removed, 60 articles were removed, leaving 53 included (44 Scopus, 9 WoS).
- Thematic qualitative content analysis with mechanical and interpretative components; theoretical frameworks include IEEE’s Ethically Aligned Design, Bostrom’s AI alignment theory, Self-Determination Theory, and constructivism.
Limitations
- Stated: quality and availability of data may restrict the comprehensiveness of the legal challenges addressed; lack of longitudinal studies hampers predicting long-term effects of current regulatory measures; rapid GenAI development leaves room for unforeseen future challenges.
- Evident: the review is limited to Scopus and Web of Science and to article-type documents, so it may miss other databases, languages, or grey literature; Table 1 mislabels GDPR and the EU AI Act as “USA,” indicating possible extraction or accuracy issues; a correction note indicates the article was corrected.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures
Verification
Claims the paper makes
18 Valid · 2 Not valid · 59 Not checked · 42 Own results and interpretations
Every source the paper cites for this claim was checked and supports it.
In the paper
“Finally, insufficient digital literacy can lead to misuse or misinterpretation of GenAI's capabilities and limits (Walczak and Cellary, 2023).”
What the sources say
- Supportedcites Walczak and Cellary, 2023 matched to “Challenges for Higher Education in the Era of Widespread Access to Generative AI” (2023) source
Section 5.1 links digital literacy to responsible AI use and to reducing the risk of accepting misleading generated content. Section 5.4 says AI literacy helps students understand AI’s strengths and limitations and make informed decisions about when and how to use it. Together, these support the claim as a paraphrase. (LLM, conf=0.90)
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
Every source the paper cites for this claim was checked and supports it.
In the paper
“Another relevant problem is academic plagiarism, where GenAI-generated content can facilitate dishonest behavior, blurring the boundaries between original effort and automated production (Williams, 2024).”
What the sources say
- Supportedcites Williams, 2024 matched to “The ethical implications of using generative chatbots in higher education” (2024) source
The Plagiarism section says AI tools may encourage students to present AI-generated outputs as their own and that chatbots can create conditions conducive to academic dishonesty. The Conclusion repeats that misrepresenting AI-generated work compromises academic integrity, supporting the paraphrase about blurring original effort and automated production. (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: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“The first of these, privacy, is at risk when personal data is mishandled, especially if generative models are trained on sensitive information without proper regulation (Williams, 2024).”
What the sources say
- Supportedcites Williams, 2024 matched to “The ethical implications of using generative chatbots in higher education” (2024) source
The Data privacy section says chatbots learn from conversations that may contain personal information and that misuse or unauthorised access to sensitive student data threatens privacy. It also calls for secure data handling and data-protection measures; the Conclusion recommends clear guidelines aligned with legislation such as GDPR. (LLM, conf=0.89)
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.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“Finally, cognitive autonomy is threatened when students rely excessively on these tools, limiting their critical thinking and autonomy in learning (Al-Kfairy et al., 2024).”
What the sources say
- Supportedcites Al-Kfairy et al., 2024 matched to “Ethical challenges and solutions of generative AI: an interdisciplinary perspective” (2024) source
Section 4.2.5, Educational Ethics, says overreliance on generative AI threatens students’ critical thinking and problem-solving skills because ready-made answers may discourage deeper engagement with learning material. This supports the claim’s characterization of a threat to autonomy in learning. (LLM, conf=0.90)
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.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“This case showed that the use of AI in education, if not designed with equity in mind, can reinforce structural inequalities rather than mitigate them (Smith, 2020).”
What the sources say
- Supportedcites Smith, 2020 matched to “Algorithmic bias: should students pay the price?” (2020) source
In the main text, Smith describes an unfair grading algorithm that favoured students at private schools while hitting students from disadvantaged backgrounds hardest. The paper warns that continued use of biased algorithms would perpetuate a growing divide between those groups, supporting the citing sentence’s paraphrase. (LLM, conf=0.91)
plainstates a possible causeSingle case- 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.
- Supported
30 more claims · 4 with their cited sources checked
Create a free account to see all 121 claims and to verify the 20 references still open.
- It may state two separate claims as one. 34
- Probably not a substantial claim. 70
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 cited source says what the claim attributes to it.
In the citing paper
“One primary concern pertains to the accuracy and reliability of AI-generated content, including introducing errors or hallucinations that could distort the learning experience, as highlighted in the landmark UNESCO report on safe and fair usage of generative AI in education and subsequent literature reviews (86–88).”
From Global Health Trade Summit: an AI-enhanced simulation of international trade and global health for undergraduate public health education (2025) · cites this paper as [88]
The cited source says what the claim attributes to it.
In the citing paper
“Research underscores the ethical complexities of AI in education, such as data privacy, algorithmic bias, AI misuse, misinformation, and disinformation, which demand clear policy, proactive institutional frameworks, and pedagogical strategies (Garcia-López & Trujillo-Liñán, 2025).”
From Policy analysis of higher education institutes in Ontario, Canada: A focus on artificial intelligence (2025) · cites this paper as Garcia-López & Trujillo-Liñán, 2025
The cited source says what the claim attributes to it.
In the citing paper
“Research underscores the ethical complexities of AI in education, such as data privacy, algorithmic bias, AI misuse, misinformation, and disinformation, which demand clear policy, proactive institutional frameworks, and pedagogical strategies (Garcia-López & Trujillo-Liñán, 2025).”
From Policy analysis of higher education institutes in Ontario, Canada: A focus on artificial intelligence (2025) · cites this paper as Garcia-López & Trujillo-Liñán, 2025
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
§ FIGURE 5 · p. 9
“| Lyu et al. (2025) | Quantitative”
§ Publisher's note · p. 13
“Understanding the practices, perceptions, and (dis)trust of generative AI among instructors: a mixedmethods study in the U.S. higher education.”
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
§ 3.1 Phase 1: research questions · p. 5
“find publications on GenAI in higher education in the period 20182023.”
§ 3.2 Phase 2: search process · p. 5
“The delimiters were keywords (generative artificial intelligence and ethics), period (2020-2024) and type of document.”
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
§ 2.5 Quality of education: building a legal framework for equity and innovation · p. 5
“AI Act (European Union) (Comisión Europea, 2021) | USA”
§ Publisher's note · p. 12
“Comisión Europea. (2021). Propuesta de Reglamento del Parlamento Europeo y del Consejo por el que se establecen normas armonizadas en materia de inteligencia artificial (Ley de Inteligencia Artificial). En Diario Oficial de la Unión Europea.”
Citations
Cites
Ryan, R. M., & Deci, E. L. (2000) · American Psychologist · doi:10.1037//0003-066x.55.1.68 · Cited by 54,599
(1978) · Medical Entomology and Zoology · Cited by 53,488
On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? 🦜
Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. (2021) · Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency · doi:10.1145/3442188.3445922 · Cited by 6,988
The Cambridge Handbook of the Learning Sciences
(2005) · Cambridge University Press eBooks · doi:10.1017/cbo9780511816833 · Cited by 4,601
Superintelligence: paths, dangers, strategies
(2015) · Choice Reviews Online · doi:10.5860/choice.187536 · Cited by 1,979
The children's machine: rethinking school in the age of the computer
(1993) · Choice Reviews Online · doi:10.5860/choice.31-1648 · Cited by 1,955
A comprehensive AI policy education framework for university teaching and learning
Chan, C. K. Y. (2023) · International Journal of Educational Technology in Higher Education · doi:10.1186/s41239-023-00408-3 · Cited by 1,577 · cited 1×
“Collaborative governance requires the joint work of politicians, educators and technologists to formulate applicable and coherent guidelines at the local and international levels (Chan, 2023).”
Artificial intelligence in education: Addressing ethical challenges in K-12 settings
Akgun, S., & Greenhow, C. (2022) · AI and Ethics · doi:10.1007/s43681-021-00096-7 · Cited by 1,473 · cited 2×
“The personalization of learning is manifested through GenAI’s ability to adapt educational content and methods to the specific needs of each student (Akgun and Greenhow, 2022; Alali and Wardat, 2024).”
Learning Analytics: The Emergence of a Discipline
Siemens, G. (2013) · American Behavioral Scientist · doi:10.1177/0002764213498851 · Cited by 1,284 · cited 1×
“Siemens (2013) highlights that educational technology is only effective when it is integrated within learning models that prioritize creativity and problem-solving over simple task automation.”
Systematic literature reviews in software engineering – A tertiary study
Kitchenham, B., Pretorius, R., Budgen, D., Pearl Brereton, O., Turner, M., Niazi, M., & Linkman, S. (2010) · Information and Software Technology · doi:10.1016/j.infsof.2010.03.006 · Cited by 1,230 · cited 2×
“The qualitative content analysis sought inferences from the information objectively considering two components, mechanical and interpretative (Kitchenham et al., 2010).”
Research works printed but not linked: 3
- Contreras, 2024, Convergencia internacional y caminos propios: Regulación de la inteligencia artificial en América Latina, Actual. Juríd. Iberoam., 21, 468
- Jonassen, 1999, Designing constructivist learning environments, 215
- Kitchenham, 2007, Guidelines for performing systematic literature reviews in software engineering (EBSE Technical Report EBSE-2007-01). Durham, United Kingdom: Department of Computer Science, University of Durham and Keele University.
Cited by
Items 101–136 of 136
Engineering Innovation: Impact of AI and SDGs in a Challenge-Based Environment
Delfa-Baena, S., Boluda-Prieto, M., Terron-Santiago, C., & Sapena-Bañó, Á. (2026) · Knowledge · doi:10.3390/knowledge6030018
“Their findings reveal strong instructor support for PBL methodologies alongside cautious optimism regarding AI, with concerns centred on infrastructure, equity, and responsible use [20–22].”
Toledo, H., & Zuloaga, S. G. (2026) · Polymathia Revista Pensamiento Académico · doi:10.33264/prpa.2026.01.17
Košuta, N., & Patekar, J. (2026) · 2018 International Conference on Multidisciplinary Research · doi:10.26803/myres.2026.08
A theoretical model for the ethical use of generative AI tools in higher education assessments
Baig, M. I., & Yadegaridehkordi, E. (2026) · International Journal of Ethics and Systems · doi:10.1108/ijoes-09-2025-0538 · Cited by 1
Petricini, T. A., Liñán, L. T., & Meneses-Trujillo, R. (2026) · Online Media and Global Communication · doi:10.1515/omgc-2026-0018
“The design was informed by COIL principles and the constructivist AI pedagogy outlined in García-López and Trujillo-Liñán (2025).”
Self-citationMante, J., Abeji, B., Kalutarage, H., Pirie, C., & Senanayake, J. (2026) · AI and Ethics · doi:10.1007/s43681-026-01278-x
Abu-Amara, F., & Hasasneh, A. (2026) · Springer international handbooks of education · doi:10.1007/978-3-032-10711-4_9-1
Designing and Developing Rights-Based Architectures for GenAI in Special and Inclusive Classrooms
Alam, A. (2026) · Advances in computational intelligence and robotics book series · doi:10.4018/979-8-3373-9169-4.ch004
Siyu, J. (2026) · Advances in Psychology · doi:10.12677/ap.2026.168382
Expected Benefits and Challenges in Using Artificial Intelligence Among Pre-Service Teachers
Cubillas, A. U., Gonzaga, J., Guinsosd, L. P., et al. (2026) · International Conference on e-Education, e-Business, e-Management and e-Learning · doi:10.1109/ic4e69618.2026.11631561
“This study used an adopted instrument from [6] called Artificial Adoption Intelligence: The Teachers' Eye Questionnaire (AAITEQ) to assess pre-service teachers' perceptions, adoption, and integration of AI in education.”
AI For Accessible and Inclusive Education in Developing and Least Developed Countries
Suresh, C., Varadharajan, D., Sunkesula, M. A., & Murathoti, R. N. B. (2026) · Advances in computational intelligence and robotics book series · doi:10.4018/979-8-3373-7225-9.ch013
Penha-Vasconcelos, C., Behar, A. I., Al‐Gasem, N. S., Silvaggio, D. N., Ridgley, L. M., & Myers, P. (2026) · New Directions for Student Leadership · doi:10.1002/yd.70074
“…time, the integration of generative AI in education (AIED) is creating both opportunities and challenges that require attention concerning ethical (Garcia-Lopez and Trujillo-Linan 2025 ; Gouseti et al. 2025 ), pedagogical (Tripathi et al. 2025 ) and leadership (Qian and Tang 2025 ; Fullan et al.…”
Arthur, F., Opoku, E., Gyedu, F. O., Quaye, S. A., Quayson, E., Boateng, E., Nortey, S. A., & Frempong, D. (2026) · Advances in computational intelligence and robotics book series · doi:10.4018/979-8-3373-7902-9.ch004
Generative AI in Education: Practical and Ethical Teaching Approaches
Barreto, D., & Colon, A. (2026) · Advances in computational intelligence and robotics book series · doi:10.4018/979-8-3373-7902-9.ch006
Aref, E., Aljabr, B., Aljasem, M., & Kluse, C. J. (2026) · ASEE Annual Conference & Exposition · doi:10.18260/1-2--58700
Dalyanci, A. A., Çelik, İ., Kontkanen, S., Laru, J., & Siklander, P. (2026) · Journal of Educational Computing Research · doi:10.1177/07356331261479569 · Cited by 2
Sneha.V.K, S. V. K. (2026) · International Journal of Technology & Emerging Research · doi:10.64823/ijter.2621019
“Additionally, excessive student reliance on automated outputs may lead to "metacognitive laziness," where AI acts as a substitute for, rather than a scaffold for, critical and creative thinking[7].”
Fenty, N. S., & Park, J. (2026) · Advances in computational intelligence and robotics book series · doi:10.4018/979-8-2600-3840-6.ch008
Non-automatable cognitive skills in higher education in the age of generative AI
Jog, S. (2026) · Frontiers in Education · doi:10.3389/feduc.2026.1827610
Rogayan, D. V., & Yektie Nurprayoga, L. M. (2026) · International Journal of Education Reports · doi:10.62503/ijer.1.1.45-58
“…and more efficient lesson preparation, yet it has also raised new concerns about academic integrity and the quality of students' thinking processes (Cotton et al., 2024; Fütterer et al., 2023; García-López & Trujillo-Liñán, 2025; Klarin et al., 2024; Kofinas et al., 2025; Yusuf et al., 2024).”
Leadership 4.0: Visionary Management in CTE and Global Higher Education
Wang, V. (2026) · Technical and vocational education and training · doi:10.1007/978-981-92-2909-3_5
Redefining Career and Technical Education (CTE): Preparing Innovators for a Disrupted World
Wang, V. (2026) · Technical and vocational education and training · doi:10.1007/978-981-92-2909-3_1
Mohammed, Y., Qandil, H., & Lund, B. (2026) · Discover Artificial Intelligence · doi:10.1007/s44163-026-02086-2
“Furthermore, related work has also examined ethical and regulatory concerns, including privacy, algorithmic bias, educational inequality, and cognitive autonomy [10], as well as the need for GenAI literacy and stronger ethics instruction rather than policy awareness alone [16, 17].”
Singh, A., Boruah, B., & Rajkhowa, P. (2026) · Journal of Psychologists and Counsellors in Schools · doi:10.1177/20556365261483215
Predictors of the ethical use of generative artificial intelligence in higher education
Cruz, Y. B. T., Sánchez, H. A. C., Pita, I. G. A., Rodríguez-Álava, L. A., Carreño-Mendoza, Á. L., Cedeño-Barreto, M., Castillo, G., Villavicencio, J. C., & Moreira-Chóez, J. S. (2026) · Frontiers in Education · doi:10.3389/feduc.2026.1942426 · Cited by 2
Velmurugan, R., Sudarvel, J., Bhuvaneswari, R., & Thirumalaisamy, R. (2026) · Advances in computational intelligence and robotics book series · doi:10.4018/979-8-2600-0383-1.ch013
Charting a human-Centered future: AI's promise and challenges for occupational therapy
Pellerin, M., Malfitano, A. P. S., Francis, J., Teng, M. Y., & Jacobs, K. (2026) · Work · doi:10.1177/10519815261490547
Basra, R. (2026) · International Journal of Modern Science and Research Technology · doi:10.67317/ijmsrt26sep073
Human-AI collaboration for ethical, transparent, and equitable educational practices
Singh, D. (2026) · Journal of Digital Educational Technology · doi:10.29333/jdet/19286
“Extending this discussion, García-López and Trujillo-Liñán (2025) demonstrated that GenAI introduces additional concerns related to privacy, bias, inequality, learner autonomy, and regulatory limitations.”
Kilwake, R. W., & Wechuli, A. N. (2026) · International Journal of Scientific Research in Science Engineering and Technology · doi:10.32628/ijsrset2613508
Ciaschi, M., & Dan, D. (2026) · Annals of Computer Science and Information Systems · doi:10.15439/2026f8742
Why Educators Use But Do Not Endorse Artificial Intelligence in Higher Education
Yu, J. H., Romero, P., Dunlap, M., & Warren, S. (2026) · Innovative Higher Education · doi:10.1007/s10755-026-09957-7
Alvero, J. C. (2026) · Journal of interdisciplinary perspectives · doi:10.69569/jip.2026.433 · Cited by 3
Urbáez, E. R. V. (2026) · Tendencias e Innovación en la Sociedad Digital TISD · doi:10.37883/tisd.2025.5.1-2.3
Andinnari, S. R., Chaeruman, U. A., & Siregar, E. (2026) · Tadris Jurnal Keguruan dan Ilmu Tarbiyah · doi:10.24042/tadris.v11i3.32309
Scifo, L., & Fulantelli, G. (2026) · Communications in computer and information science · doi:10.1007/978-3-032-31850-3_7
Metadata
- Authors
- García-López, I. M., & Trujillo-Liñán, L.
- Corresponding author
- García-López, Iván Miguel
- Venue
- Frontiers in Education
- Publisher
- Frontiers Media SA
- Published
- 2025-06-30
- Volume
- 10
- ISSN
- 2504-284X (ISSN-L)
- OpenAlex
- W4411842024
- Cited by
- 144
- References
- 48
- Citations by year
- 2026: 117 · 2025: 27
- FWCI
- 21.94
- Type
- journal article
- Language
- en
- Update notices
- correction (2025-09-08)
- Open access
- gold
- doi.org
- Licence
- Locations
- Frontiers in Education (publishedVersion, cc-by)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Topics
- Social Sciences › Education › Artificial Intelligence in Education
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
- Keywords
- generative AI, artificial intelligence in education, ethical considerations, regulatory framework, data privacy, algorithmic bias, educational inequality, personalized learning, cognitive autonomy, digital literacy, self-determination theory, constructivist learning, higher education, responsible AI adoption
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 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (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.




