Contents
Academy · Research Publication · AI in Scholarly Publishing
The Role of AI in Scholarly Publishing: Authorship, Disclosure and Publisher Policies
Within weeks of the release of ChatGPT in late 2022, several published papers listed it as a co-author before major publishers intervened. The International Committee of Medical Journal Editors (ICMJE), the World Association of Medical Editors (WAME) and many publishers have since issued guidance stating that AI tools may assist with writing but cannot be authors. Authors should check the target journal’s current policy on AI-assisted work and disclose its use where required.
Generative AI tools entered scholarly publishing abruptly rather than gradually, and publishers' policies were obliged to respond to practices already under way. Within months of large language models becoming widely accessible in late 2022, researchers submitted and, in a handful of documented cases, successfully published papers listing an AI tool as a co-author, prompting an immediate and largely unified response from the major bodies that set publication norms. The resulting consensus, now reflected in ICMJE's recommendations, COPE's position statement and individual publisher policies, draws a specific distinction: AI assistance in writing or analysis must be disclosed, but an AI tool cannot meet the accountability standard that authorship requires.
Learning objectives
- The reasons AI tools cannot meet the authorship criteria of ICMJE, as stated explicitly in its guidance
- The disclosure requirements and prohibitions set out in the COPE and WAME position statements
- The operationalization of these principles in the submission requirements of individual publishers
Common misconceptions
- “If an AI tool substantially drafted part of a paper, listing it as a co-author is just accurate attribution.” ICMJE's updated guidance states explicitly that AI tools cannot meet authorship requirements because they cannot take responsibility for the submitted work, cannot agree to be accountable for its accuracy, and cannot approve a final version in any meaningful legal or ethical sense, which rules out AI co-authorship regardless of how much of the drafting it performed.
Ineligibility of AI tools for authorship
ICMJE's updated recommendations address AI-assisted technology directly, stating that such tools do not meet the requirements for authorship because they cannot be responsible for the accuracy, integrity, or originality of the work, cannot agree to be accountable for the paper's content if questions arise, and cannot grant or deny permission for the final version to be published1. This follows directly from the same four-part authorship standard that ICMJE applies to human contributors: accountability is not incidental to authorship but is the function that authorship exists to assign, and a tool with no legal or professional standing cannot hold it1. The World Association of Medical Editors' recommendations on chatbots and generative AI reach the same conclusion independently, adding that authors remain fully responsible for any content generated with AI assistance, including for checking it against primary sources, because the tool's output cannot be assumed to be accurate or original by default2.
Disclosure requirements and their placement
Where ICMJE and WAME converge on prohibiting AI authorship, both bodies converge equally firmly on requiring disclosure of AI assistance that does occur. ICMJE recommends that authors who used AI-assisted technology in writing, editing, image creation, or data analysis for a submission describe how it was used, in the manuscript itself, typically in the methods or acknowledgments section, with enough specificity that an editor or reader can understand the tool's actual role1. COPE's 2023 position statement on authorship and AI tools frames this as a transparency obligation that exists independently of whether the AI's contribution was substantial: even relatively minor use, such as AI-assisted language polishing, should be disclosed so that readers are not misled about how the manuscript was produced3. Both bodies also address the reverse direction, AI use in peer review, cautioning that confidential, unpublished manuscript content should not be uploaded to AI tools that may retain or reuse submitted data, since doing so can breach the confidentiality obligations peer review depends on3.
Implementation of AI policies by individual publishers
Major publishers moved quickly to translate these principles into submission-system requirements. Nature's editorial position, published as a 2023 editorial, states that large language models do not meet its authorship criteria and will not be accepted as credited authors on any paper, while requiring that any use of such tools be documented in the paper's methods section4. The same editorial followed closely after Nature's own reporting documented specific instances of ChatGPT being listed as a co-author on at least four preprints or papers in the weeks after its public release, the triggering events that prompted the policy response5. Elsevier's and Springer Nature's publicly stated author policies both mirror the same two-part structure: an outright prohibition on listing AI tools as authors, paired with a mandatory disclosure requirement wherever generative AI materially contributed to the manuscript's text, figures or analysis6. For authors, the practical implication is currently the same across every major publisher: the use of AI tools in writing or analysis is permitted and increasingly common, but it must be disclosed specifically, and it never substitutes for the human accountability that authorship requires.
Applications of AI beyond manuscript drafting
Generative AI's role in scholarly publishing extends past manuscript drafting into peer review, editorial triage, and post-publication screening, each raising distinct concerns the writing-focused disclosure rules do not fully address. On the review side, WAME's recommendations specifically caution reviewers against using AI tools to draft or summarize their review of a confidential, unpublished manuscript, both because of the data-confidentiality risk already noted and because an AI-generated review risks missing the specific, field-expert judgment peer review exists to provide, substituting a generic assessment for a substantive one2. On the editorial and integrity side, publishers have begun using AI-based tools defensively, to screen submissions for signs of AI-generated text that was not disclosed, for image manipulation, and for the structural patterns associated with paper-mill production, treating detection as a distinct, sanctioned use case separate from AI-assisted authoring3. This creates a notable asymmetry: the same class of tool is being restricted on the authoring side (disclosure required, authorship prohibited) while being actively adopted on the integrity-screening side, a distinction that follows from the same underlying principle, that a tool without accountability should not make or be credited with substantive judgments, but can legitimately assist human reviewers and editors who remain accountable for the final decision1.
Common errors in AI-use disclosure
In practice, authors trying to comply with these rules make a few specific, recurring mistakes. The first is disclosing AI use too generically, a single blanket sentence stating that "AI tools were used in preparing this manuscript" without naming the specific tool or its specific task does not meet ICMJE's or COPE's actual recommendation, which calls for enough detail that an editor or reader can understand what the tool actually did1. The second is treating AI-generated citations or factual claims as reliable without independent verification; large language models have a well-documented tendency to generate plausible but nonexistent citations or to misstate details of real sources, which is why WAME's guidance places the verification burden explicitly on the human author rather than treating disclosure alone as sufficient diligence2. The third is assuming a single disclosure statement covers every stage of a project, when in fact drafting assistance, data-analysis assistance, and image-generation assistance are typically expected to be disclosed separately and specifically, since each raises distinct verification and originality concerns that a single combined statement tends to obscure. Treating AI disclosure with the same specificity expected of a methods section, naming the tool, the version where relevant, the specific task, and the verification step taken, is the practical standard that satisfies every major publisher's current policy simultaneously, rather than requiring a different disclosure statement tailored to each individual journal's particular wording.
Anticipated revision of AI policies
The speed with which ICMJE, WAME, COPE, and individual publishers converged on their current position after late 2022 is itself a signal that these specific policies should be read as a current snapshot rather than a permanently settled standard. All four of the governing documents cited throughout this topic, ICMJE's recommendations, WAME's chatbot guidance, COPE's position statement, and individual publisher policies, have already been revised at least once since their initial publication as the underlying technology and its observed use in submissions have changed, and each body has signaled it expects to continue revising its guidance as new capabilities and new failure modes emerge13. For an author or institution relying on this guidance, the practical implication is to check the specific, current version of a target journal's AI policy at the time of submission rather than assuming a policy noted at an earlier point, even a recent one, is still current, since the underlying consensus on prohibiting AI authorship has held steady while the specific disclosure mechanics and permitted-use boundaries around it have continued to be refined. This holds equally for authors reusing a disclosure statement drafted for an earlier submission; copying it forward without checking whether the specific tool, version, or task it describes still matches the current manuscript, or whether the target journal's policy has since been updated, is itself a documented, avoidable source of inaccurate disclosure.
Key points
Key points
AI tools cannot hold authorship under the standard of any major body because they cannot be accountable for the accuracy of a paper. Every major publisher now permits disclosed use of AI in writing or analysis; the practical obligation for authors is therefore specific, written disclosure rather than avoidance of the tools.
Practical application
Practical application
- An AI tool should not be listed as a co-author, regardless of the scale of its contribution. Substantial AI assistance should be credited through disclosure in the methods or acknowledgments, not through the byline.
- AI use should be disclosed specifically rather than generically. Authors should state which tool was used and for which task (drafting, editing, analysis, image generation), and should verify every AI-generated claim against a primary source before submission.
- Unpublished manuscript content should not be uploaded to a public AI tool during peer review. Reviewers should establish whether a tool retains or reuses submitted data before using it on any confidential material under review.
Frequently asked questions
No major publisher or standards body currently allows this. ICMJE, WAME, COPE, and individual publishers including Nature, Elsevier, and Springer Nature all state that AI tools cannot meet authorship criteria because they cannot be held accountable for the work.
COPE and ICMJE both recommend disclosing which tool was used and for what specific purpose (drafting, editing, analysis, etc.), typically in the methods or acknowledgments section, regardless of how substantial the contribution was.
WAME and COPE both caution against uploading unpublished, confidential manuscript content to AI tools that may retain or reuse submitted data, since this can breach the confidentiality obligations of the review process.
Nature's own reporting documented AI tools being listed as co-authors on at least four papers or preprints within weeks of ChatGPT's public release, which directly triggered the rapid, largely unified publisher response that followed.
No major guidance treats disclosed AI assistance as a credibility problem; the integrity issue is non-disclosure or unverified AI-generated content, not the use of the tool itself when properly documented and checked.
References
- International Committee of Medical Journal Editors. Recommendations: Artificial Intelligence (AI)-Assisted Technology. read 10 October 2026. View on ICMJE
- World Association of Medical Editors. Chatbots, Generative AI, and Scholarly Manuscripts: WAME Recommendations. read 10 October 2026. View on WAME
- Committee on Publication Ethics. Authorship and AI Tools: COPE Position Statement. read 10 October 2026. View on COPE
- Nature Editorial. Tools such as ChatGPT threaten transparent science; here are our ground rules for their use. Nature. 2023;613:612. read 10 October 2026. View on Nature
- Stokel-Walker C. ChatGPT listed as author on research papers. Nature. 2023;613:620-621. read 10 October 2026. View on Nature
- Elsevier. The Use of AI and AI-assisted Technologies in Writing for Elsevier. read 10 October 2026. View on Elsevier
- International Committee of Medical Journal Editors. Recommendations: Artificial Intelligence (AI)-Assisted Technology. read 10 October 2026. View on ICMJE
- World Association of Medical Editors. Chatbots, Generative AI, and Scholarly Manuscripts: WAME Recommendations. read 10 October 2026. View on WAME
- Committee on Publication Ethics. Authorship and AI Tools: COPE Position Statement. read 10 October 2026. View on COPE
- Stokel-Walker C. ChatGPT listed as author on research papers. Nature. 2023;613:620-621. read 10 October 2026. View on Nature
Every source above is a peer-reviewed article, a publisher’s record of one, or a standard issued by the body that owns it. Each carries the date it was last read. Where a standard moves, this article is revised rather than left up with a note.
How to cite this article
Pick your referencing style and copy it. Your university or target journal may vary the punctuation — check its guide if it has one.
Pubrica Academy. (2026, October 10). The Role of AI in Scholarly Publishing: Authorship, Disclosure and Publisher Policies. Pubrica. https://pubrica.com/academy/academy-research-writing/the-role-of-ai-in-scholarly-publishing-2/
Pubrica Academy. “The Role of AI in Scholarly Publishing: Authorship, Disclosure and Publisher Policies.” Pubrica, 10 Oct. 2026, pubrica.com/academy/academy-research-writing/the-role-of-ai-in-scholarly-publishing-2/.
Pubrica Academy (2026) The Role of AI in Scholarly Publishing: Authorship, Disclosure and Publisher Policies. Pubrica. Available at: https://pubrica.com/academy/academy-research-writing/the-role-of-ai-in-scholarly-publishing-2/ (Accessed: 15 October 2026).
Pubrica Academy. “The Role of AI in Scholarly Publishing: Authorship, Disclosure and Publisher Policies.” Pubrica. October 10, 2026. https://pubrica.com/academy/academy-research-writing/the-role-of-ai-in-scholarly-publishing-2/
Pubrica Academy, ‘The Role of AI in Scholarly Publishing: Authorship, Disclosure and Publisher Policies’ (Pubrica, 10 October 2026) <https://pubrica.com/academy/academy-research-writing/the-role-of-ai-in-scholarly-publishing-2/> accessed 15 October 2026.
Pubrica Academy. The Role of AI in Scholarly Publishing: Authorship, Disclosure and Publisher Policies. [Internet]. Pubrica; 2026 Oct 10 [cited 2026 Oct 15]. Available from: https://pubrica.com/academy/academy-research-writing/the-role-of-ai-in-scholarly-publishing-2/
Pubrica Academy, “The Role of AI in Scholarly Publishing: Authorship, Disclosure and Publisher Policies,” Pubrica, Oct. 10, 2026. [Online]. Available: https://pubrica.com/academy/academy-research-writing/the-role-of-ai-in-scholarly-publishing-2/
@misc{/la_ai_role_scholarly_publishing_2026,
author = {{Pubrica Academy}},
title = {The Role of AI in Scholarly Publishing: Authorship, Disclosure and Publisher Policies},
howpublished = {Pubrica},
year = {2026},
month = {oct},
url = {https://pubrica.com/academy/academy-research-writing/the-role-of-ai-in-scholarly-publishing-2/},
note = {Accessed: <span data-cite-acc="iso">2026-10-15</span>}
}This article is written to be learned from. Cite it if you are referring to what it says; if you are using the evidence behind it, cite the sources listed above instead.
AI Use Disclosure Statement
This language may be adapted for the methods or acknowledgments section when AI tools have been used.
During the preparation of this work, the author(s) used [name of tool] for [specific purpose: e.g., language editing, literature search assistance, data visualization]. After using this tool, the author(s) reviewed and edited the content as needed and take full responsibility for the content of the published article. [ ] Tool named specifically [ ] Purpose of use stated specifically (not generic) [ ] Confirmed no confidential/unpublished third-party content was uploaded to the tool [ ] Verified all AI-assisted claims against primary sources
Pubrica AcademyWritten and maintained by Pubrica’s subject matter experts. Articles are revised when the underlying standards change, and each article states the date on which it was last reviewed.
How Pubrica can help
Pubrica's manuscript preparation team helps authors draft AI-use disclosure statements that meet ICMJE, WAME, and specific publisher requirements, and independently verifies AI-assisted content against primary sources before submission.