The DOI matches its Crossref record; 57 of the 58 references it cites were found in a registry or an index, and 1 are not checked yet; 3 of the 29 sources its claims cite were checked for support.
Academic publisher guidelines on AI usage: A ChatGPT supported thematic analysis
Open accessAbstract
Background As Artificial Intelligence (AI) technologies such as Generative AI (GenAI) have become more common in academic settings, it is necessary to examine how these tools interact with issues of authorship, academic integrity, and research methodologies. The current landscape lacks cohesive policies and guidelines for regulating AI’s role in academic research which has prompted discussions among publishers, authors, and institutions. Methods This study employs inductive thematic analysis to explore publisher policies regarding AI-assisted authorship and academic work. Our methods involved a two-fold analysis using both AI-assisted and traditional unassisted techniques to examine the available policies from leading academic publishers and other publishing or academic entities. The framework was designed to offer multiple perspectives, harnessing the strengths of AI for pattern recognition while leveraging human expertise for nuanced interpretation. The results of these two analyses are combined to form the final themes. Results Our findings indicate six overall themes, three of which were independently identified in both the AI-assisted and unassisted, manual analysis using common software tools. A broad consensus appears among publishers that human authorship remains paramount and that the use of GenAI tools is permissible but must be disclosed. However, GenAI tools are increasingly acknowledged for their supportive roles, including text generation and data analysis. The study also discusses the inherent limitations and biases of AI-assisted analysis, necessitating rigorous scrutiny by authors, reviewers, and editors. Conclusions There is a growing recognition of AI’s role as a valuable auxiliary tool in academic research, but one that comes with caveats pertaining to integrity, accountability, and interpretive limitations. This study used a novel analysis supported by GenAI tools to identify themes emerging in the policy landscape, underscoring the need for an informed, flexible approach to policy formulation that can adapt to the rapidly evolving landscape of AI technologies.
Verified 2026-11-13 · 1 reference not checked yet
Overview
CoChat Summary
TL;DR
This study analyzes 28 unique academic publisher AI policies using manual thematic analysis and ChatGPT-assisted analysis, finding broad consensus that human authorship remains paramount and GenAI use is allowed but must be disclosed. It also demonstrates a dual-method approach, with three themes identified independently by both analyses and AI-generated quotes requiring verification due to hallucinations.
Key findings
- From 107 entities reviewed, 36 had AI policies; after removing 8 replicas, 28 unique policies were analyzed. Six final themes emerged: Human-Exclusive Authorship, Author Accountability, Disclosure and Transparency, Research Integrity, Fluid Policy Landscape, and Constraints and Exclusions; three were independently identified in both AI-assisted and manual analyses.
- Publishers broadly agree AI tools cannot be listed as authors because they cannot assume ethical or legal responsibility, and authors remain accountable for all content, including AI-assisted or generated portions.
- GenAI use is generally permitted for support such as text generation and data analysis, but disclosure is a near-universal requirement, though the degree and format of disclosure vary; some policies impose constraints or outright prohibitions.
- No significant differences in approach were noted across publisher categories, and the AI-assisted analysis produced inaccurate, misattributed, condensed, or fabricated policy quotes that required manual verification.
Methods
- Inductive thematic analysis following Braun & Clarke’s six-phase approach; manual analysis by Researcher 2 using QDA Miner, and AI-assisted analysis by Researcher 1 using ChatGPT Plus/GPT-4 default and Advanced Data Analysis modes on July 27 and September 11, 2023, with results combined into final themes.
- A master list of 107 entities was compiled from publisher rankings/lists and web searches, excluding suspected predatory publishers; manual website searches found 36 entities with AI policies, 8 replicas, leaving 28 unique policies.
- Coding was at the semantic level; final themes were developed by comparing and grouping manual and AI-assisted themes, with manual editing and verification of quotes.
Limitations
- Stated: the analysis was confined to a limited set of publisher policies available at the time; it focused on English-language and flagship/major publishing houses, excluding non-academic publishing and professional/governmental guidelines; it used a single AI tool, though four analyses over two time periods; GenAI may have missed nuanced interpretations while human analysis was limited by time and cognitive load; and the study did not empirically test the effectiveness or reliability of AI tools.
- Stated: the data were straightforward and simple compared with interview data, so further studies are needed, and findings should be considered a snapshot of an evolving landscape.
- Evident from the text: the AI-assisted analysis hallucinated, misattributed, condensed, or fabricated policy quotes, and only 28 policies were analyzed, so coverage may be incomplete and rapidly outdated.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
3 Valid · 2 Not valid · 21 Not checked · 152 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Despite the widespread use of digital and AI technologies, only a small percentage of higher education institutions (HEIs) have developed formal policies surrounding their use since the launch of ChatGPT ( Perkins & Roe, 2023; Xiao et al., 2023).”
§ Academic publisher guidelines on AI usage: A ChatGPT supported thematic analysis > Introduction
What the sources say
- Not matchedcites Perkins & Roe, 2023
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Partly supportedcites Xiao et al., 2023 sourceExperimental — accuracy not yet measured
Xiao et al. (2023) do report that only a small share of institutions had issued such policies — the Introduction (p. 1) states 'less than a third (26.53%) of the top 500 universities have responded with a ChatGPT [policy]', and Section 2 gives 132 of the top 500 (26.35%), with the Abstract and Discussion repeating 'less than one-third'. However, that figure is bounded to the 500 highest-ranked universities in the 2022 QS World University Rankings, a selective, reputation- and English-speaking-country-skewed sample; the paper nowhere estimates a rate across higher education institutions generally, so the citing sentence's extension to 'higher education institutions (HEIs)' at large is not supported by this source. The source's policies are also specifically ChatGPT/generative-AI policies identified by searching each university's official website, rather than formal GenAI policy in general. (LLM, conf=0.78)
plain- 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.
- Not matched
Every source the paper cites for this claim was checked and supports it.
In the paper
“Even advanced AI detection software tools struggle, with a detection rate of only 54.8% in examples of work submitted to the Turnitin platform ( Perkins et al., 2023), and inaccuracies in a range of detection tools have been identified and discussed by Weber-Wulff et al. (2023).”
§ Academic publisher guidelines on AI usage: A ChatGPT supported thematic analysis > Literature review > AI’s role and ethical implications in academic research
What the sources say
- Supportedcites Weber-Wulff et al., 2023 source
The Discussion (p. 25) states outright that 'Detection tools for AI-generated text do fail, they are neither accurate nor reliable (all scored below 80% of accuracy and only 5 over 70%)', with both false positives and false negatives across the 14 tools tested (tool list and selection described p. 9). The results sections quantify these inaccuracies per tool (Table 14 / p. 21 false-negative rates; pp. 20-21 false accusations), and the Conclusion (p. 27) summarises the 'serious limitations' of the tools, so the attribution of identified and discussed inaccuracies in a range of detection tools to this paper is supported. (LLM, conf=0.92)
plain- 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
“However, these guidelines can only go so far, given that accurately identifying text produced by tools such as ChatGPT remains a major challenge ( Hu, 2023; Weber-Wulff et al., 2023; Perkins, Roe et al., 2023; Anderson et al., 2023).”
§ Academic publisher guidelines on AI usage: A ChatGPT supported thematic analysis > Literature review > AI’s role and ethical implications in academic research
What the sources say
- Not matchedcites Hu, 2023
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Supportedcites Weber-Wulff et al., 2023 source
The Related work section (p. 8) states outright that 'Detection of the AI-generated text remains challenging for existing detection tools', noting accuracy around 50% for AI-generated text versus above 80% for human-written text. The Discussion (p. 25) reinforces this with the paper's own findings: detection tools 'do fail, they are neither accurate nor reliable (all scored below 80% of accuracy and only 5 over 70%)', with a main bias towards classifying output as human-written. The hedged/broad framing of the citing sentence is therefore fully carried by this source. (LLM, conf=0.92)
- Not matchedcites Perkins, Roe et al., 2023
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
plain- 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.
- Not matched
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Ambiguity in this sense does not cover whether it is unacceptable or acceptable to use a GenAI tool, as it is implied and explicitly stated in several policies that their use is allowed in the preparation of manuscripts for publication.”
§ Academic publisher guidelines on AI usage: A ChatGPT supported thematic analysis > Results > Results section one: Manually generated thematic analysis
plain- 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.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Although it may seem that publisher policy documents would be formulaic and straightforward, with little room for deeper analysis, three key themes were identified from the data.”
§ Academic publisher guidelines on AI usage: A ChatGPT supported thematic analysis > Results > Results section one: Manually generated thematic analysis
plain- 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.
- Rests on an assumption not shown here.
49 more claims
Create a free account to see all 178 claims and to verify the 13 references still open.
- Probably not a substantial claim. 100
- It may state two separate claims as one. 57
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
§ Academic publisher guidelines on AI usage: A ChatGPT supported thematic analysis > Results > Results section two: GenAI supported thematic analysis
“| Policy Evolution | _ | Impact of Evolving AI | Evolving Policies | Adaptability of Policies | Indicates evolving policies due to rapid AI advancements. |”
§ Academic publisher guidelines on AI usage: A ChatGPT supported thematic analysis > Results > Results section two: GenAI supported thematic analysis
“| Ethical Considerations | Mentions or refers to ethical guidelines and frameworks like COPE, focusing on the need for AI tools to meet ethical standards. | Ethical Considerations | Addresses ethical considerations such as plagiarism and research integrity. |”
Cited works that were retracted
Works this paper cites that were later retracted, from the retraction notices publishers deposit. Major when the notice came out before this paper was published.
A note, not an issue: for example, a cited work that was retracted only after this paper was published, which its authors could not have known.
In the paper
“Y Kumar, 2023, Artificial intelligence in disease diagnosis: A systematic literature review, synthesizing framework and future research agenda., J. Ambient. Intell. Humaniz. Comput., 14, 8459-8486, 10.1007/s12652-021-03612-z”
Cited work 10.1007/s12652-021-03612-z · retracted 2026-03-09 · notice 10.1007/s12652-026-05063-w · this paper published 2024-01-16
Citations
Cites
1 retracted
Kumar, Y., Koul, A., Singla, R., & Ijaz, M. F. (2023) · Journal of Ambient Intelligence and Humanized Computing · doi:10.1007/s12652-021-03612-z · Cited by 1,040
RetractedUsing thematic analysis in psychology
Braun, V., & Clarke, V. (2006) · Qualitative Research in Psychology · doi:10.1191/1478088706qp063oa · Cited by 194,414
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
Dwivedi, Y. K., Kshetri, N., Hughes, L., Slade, E. L., Jeyaraj, A., Kar, A. K., Baabdullah, A. M., Koohang, A., Raghavan, V., Ahuja, M., et al. (2023) · International Journal of Information Management · doi:10.1016/j.ijinfomgt.2023.102642 · Cited by 4,584
BAİDOO-ANU, D., & OWUSU ANSAH, L. (2023) · Journal of AI · doi:10.61969/jai.1337500 · Cited by 3,939
Chatting and cheating: Ensuring academic integrity in the era of ChatGPT
Cotton, D. R. E., Cotton, P. A., & Shipway, J. R. (2024) · Innovations in Education and Teaching International · doi:10.1080/14703297.2023.2190148 · Cited by 3,087
Chatting and Cheating. Ensuring academic integrity in the era of ChatGPT
Cotton, D., Cotton, P., & Shipway, J. R. (2023) · doi:10.35542/osf.io/mrz8h · Cited by 3,079
Baidoo-Anu, D., & Owusu Ansah, L. (2023) · doi:10.2139/ssrn.4337484 · Cited by 3,004
Risko, E. F., & Gilbert, S. J. (2016) · Trends in Cognitive Sciences · doi:10.1016/j.tics.2016.07.002 · Cited by 1,502 · cited 1×
“On the one hand, these tools may accelerate the research process by supporting in cognitive offloading (Perkins et al. , 2023; Dawson (2022) as cited in Sparrow, 2022); the process of reducing the cognitive demands of a task (Risko & Gilbert, 2016).”
Gao, C. A., Howard, F. M., Markov, N. S., Dyer, E. C., Ramesh, S., Luo, Y., & Pearson, A. T. (2022) · doi:10.1101/2022.12.23.521610 · Cited by 1,099
Research works printed but not linked: 4
- 2023, February 20, What is SSRN’s policy on the use of generative AI and AI-assisted technologies in papers posted at SSRN? - SSRN Support Center.
- J Henley, 2023, June 20, German tabloid Bild cuts 200 jobs and says some roles will be replaced by AI., The Guardian.
- L Nowell, 2017, Thematic analysis: Striving to meet the trustworthiness criteria., Int. J. Qual. Methods., 16, 10.1177/16094069177338 · DOI
- 2023, Rankings of academic publishers., Wikipedia.
Cited by
Items 101–125 of 125
Fagroud, K., & Mrah, I. (2025) · Innovation, Technologies, Education et Communication · doi:10.60590/prsm.itec-iss10.173
Дадзи, И. (2025) · Векторы благополучия · doi:10.18799/26584956/2025/4/2045
Generative‐AI policies vary widely across top ornithology journals
McClure, C. J. W. (2026) · Ibis · doi:10.1111/ibi.70054
Selina, R. R. (2026) · Sormovo Readings · doi:10.31483/r-153964
Singh, J. P., & Kaur, B. (2026) · IGI Global eBooks · doi:10.4018/979-8-3693-8302-5.ch014
del Giglio, A. (2026) · Clinical Oncology Letters · doi:10.4322/col.2026.003
Chong, D. Y. K., Capio, C. M., JONES, A. Y. M., Tse, P. P. S., Chau, S. H. Y., & Eguia, K. F. (2026) · Physiotherapy Theory and Practice · doi:10.1080/09593985.2026.2656345
Zulkarnain, Z., Mahfud, T., Karim, S., Amsal, A. Y., Aditya, A. W., Mahsun, M., Kusuma, B. J., Kango, R., & Gozali, G. (2025) · Jurnal Hasil Inovasi Masyarakat · doi:10.70310/jhim.2025.03021284
Generative AI in Academic Publishing: Comparative Analysis of Five Publishers' Policies
Kumar, A., McKeown, D., Syed, H., Craig, C. J., Rupley, W. H., Ahmed, A., Kashwani, S., & Siddiqui, K. A. (2026) · Advances in computational intelligence and robotics book series · doi:10.4018/407619
A qualitative programme evaluation of New Era, a mentorship programme for women leaders in sport
Fortnum, K., Almarcha, M., Harrison, Z., Fitzsimmons, T. W., Richard, V., Mulcahy, S., Hinton, S., & Cairney, J. (2026) · Frontiers in Sports and Active Living · doi:10.3389/fspor.2026.1813783
Lilly, J. M., Sandoval-Cortes, L. F., & Csomor, M. (2026) · Health Communication · doi:10.1080/10410236.2026.2689188
Adamo, S. (2026) · Филолог – часопис за језик књижевност и културу · doi:10.21618/fil2633607a
The Ethics of Using Artificial Intelligence in Qualitative Research
Bogdanov, T. V. (2026) · Inter · doi:10.19181/inter.2026.18.2.4
GenAI and Authorship in Academia: The (Im)Possible Creator. An Integrative Review
Vassileva, I. G., & Chankova, M. (2026) · Discourses on Culture · doi:10.2478/doc-2026-0004
Navigating Boundaries of AI Use in Physical Education Journal Research Publications
Li, Y., Wu, C., Su, Y., Lu, B., Hamilton, X. D., Lynch, T., & Hamilton, M. F. (2026) · Quest · doi:10.1080/00336297.2026.2707164
Tehrani, M. A., Hojati, S. M., Arshadi, H., & Erfanmanesh, M. (2026) · Medical Journal of the Islamic Republic of Iran · doi:10.47176/mjiri.40.47
Cavalcanti, M. L. X. D. H., Silva, L. C. E., Costa, L., Filho, J. L. E. S., Silva, L. G. D. O., & dos Santos, F. S. (2026) · Smart Grids and Sustainable Energy · doi:10.1007/s40866-026-00363-8
der Westhuizen, A. V. (2026) · Elsevier eBooks · doi:10.1016/b978-0-443-36775-5.00003-5
Valentine Joseph Owan, P. D. (2026) · Forum qualitative Sozialforschung · doi:10.17169/fqs-27.3.4668
Sperl, M. F. J., Baumgärtner, L., Bach, K. M., Bamberg, C., Behlau, C., Bergmann, B., Bienefeld, M., Bleckmann, E., Danböck, S. K., Dreston, J. H., et al. (2026) · Psychologische Rundschau · doi:10.1026/0033-3042/a000779
Generative AI and Scientific Publishing: Preserving Accountability, Trust, and Governance
Jiang, Y., & Rathmell, J. P. (2026) · Anesthesiology · doi:10.1097/aln.0000000000006268
Andonotopo, W., Pramono, M. B. A., Dewantiningrum, J., Suryawan, A., Putra, R. A., Rahardjo, T. M., Rumanti, R. T., Lengkey, R. F., Setiawan, A., Gunadi, J. W., et al. (2026) · West Kazakhstan Public Health Journal · doi:10.4103/wkpj.wkpj_44_26
Artificial Intelligence and the Transformation of the Scientific Method
Sołoducho-Pelc, L. M., & Sulich, A. (2026) · Procedia Computer Science · doi:10.1016/j.procs.2026.09.444
Generative AI and Scientific Publishing: Preserving Accountability, Trust, and Governance
Jiang, Y., & Rathmell, J. P. (2026) · Anesthesiology Open · doi:10.1097/ao9.0000000000000034
Ahmed, A. A., Mohamed, A. A., Ali, M. K., Mohamed, Y. H., Omar, M. A., & MOHAMED, A. A. (2026) · Frontiers in Research Metrics and Analytics · doi:10.3389/frma.2026.1965518
Metadata
- Authors
- Perkins, M., & Roe, J.
- Affiliations
- British University Vietnam (VN)
- James Cook University Singapore (SG)
- Corresponding author
- Perkins, Mike
- ORCID
- Perkins, Mike
- Venue
- F1000Research
- Publisher
- F1000 Research Ltd
- Published
- 2024-01-16
- Volume
- 12
- Pages
- 1398
- ISSN
- 2046-1402 (ISSN-L)
- PMC
- PMC10844801
- OpenAlex
- W4387881071
- Semantic Scholar
- a0968924129aee94ca2070004e657df3d34a41dd
- Cited by
- 123
- References
- 53
- Citations by year
- 2026: 44 · 2025: 59 · 2024: 20
- FWCI
- 35.04
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- cc-by
- creativecommons.org/licenses/by/4.0/ (tdm)
- Locations
- F1000Research (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- ResearchOnline - JCU (James Cook University) (submittedVersion, cc-by)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Durham Research Online (Durham University) (submittedVersion)
- Topics
- Social Sciences › Education › Academic integrity and plagiarism
- Social Sciences › Library and Information Sciences › Publishing and Scholarly Communication
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Keywords
- generative AI, scholarly publishing, publisher policies, AI-assisted writing, authorship, academic integrity, AI disclosure, AI-assisted research, ChatGPT, algorithmic bias, research accountability
- MeSH terms
- Data Analysis, Artificial Intelligence, Authorship, Humans, Organizations, Software
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 2026-11-13, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-11-13 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (qwen3.8-omni-flash).
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.
