Contents
Academy · Research Publication · AI Authorship Policy
The Evolution of AI Authorship Policies in Academic Publishing
Within weeks of the release of ChatGPT in late 2022, at least four papers had listed it as a co-author. The International Committee of Medical Journal Editors (ICMJE), the Committee on Publication Ethics (COPE) and the World Association of Medical Editors (WAME), together with publishers such as Nature, Science and Elsevier, have since issued guidance stating that a generative AI tool cannot be an author. Authors should check the target journal’s current policy on AI-assisted work and disclose its use where required.
An AI authorship policy is a journal or publisher's formal rule governing whether, and how, a generative AI tool such as ChatGPT can be credited in connection with a manuscript. Such policies were introduced by editorial bodies and publishers after instances surfaced of papers listing ChatGPT itself as a named co-author, and those discussed here state that an AI tool cannot meet the criteria for authorship and that its use in preparing a manuscript must be disclosed. Their details differ and continue to be revised, so authors should check the target journal’s current policy on AI-assisted work and disclose its use where required. An understanding of what the resulting disclosure requirements ask of authors is now a practical necessity for any researcher using these tools in writing.
Learning objectives
- Why early instances of AI tools listed as co-authors prompted a rapid, coordinated editorial response
- The specific reasoning ICMJE, COPE, and WAME each gave for why an AI tool cannot be an author
- The details that disclosure requirements ask authors to specify beyond a blanket acknowledgment
Common misconceptions
- “Different journals and publishers each worked out their own, inconsistent rules on AI and authorship.” ICMJE's recommendations, COPE's position statement, the World Association of Medical Editors' (WAME) guidance, and individual policies from Nature, Science, and Elsevier share a core position: AI tools cannot satisfy authorship criteria because they cannot take accountability for the work, and their use must be disclosed in the manuscript. Implementation details still differ, so authors should check the target journal’s current policy on AI-assisted work and disclose its use where required.
Events prompting a rapid editorial response in late 2022
ChatGPT was released to the public in November 2022, and within weeks, instances surfaced of manuscripts and preprints listing the tool itself as a named co-author, a practice that immediately exposed a gap in existing authorship policy, since no journal's guidelines had anticipated a non-human entity being proposed as a credited author5. Holden Thorp, editor-in-chief of the Science family of journals, responded with an editorial in January 2023 titled "ChatGPT is fun, but not an author," announcing that Science's editorial policies would be revised to explicitly state that text generated by AI, including ChatGPT, could not be used in submitted work and that an AI tool could not be listed as an author4. The same period of concern also produced the Gao study's demonstration that fabricated, AI-generated abstracts could pass undetected through peer review and plagiarism screening, which reinforced for editors that AI's role in a manuscript needed its own dedicated policy rather than being treated as an extension of existing authorship or plagiarism rules10, a finding Nature's own news coverage treated as independent confirmation that the concern driving the new authorship and disclosure policies was concrete rather than speculative6. The speed of this response reflects the urgency with which the authorship question entered editorial agendas: this was not a gradual policy debate but a reactive measure to specific, already-published instances of AI co-authorship that required attention before further instances occurred.
Convergence of major bodies on the exclusion of AI from authorship
The International Committee of Medical Journal Editors updated its Recommendations to state explicitly that AI tools do not meet ICMJE's authorship criteria, because those criteria require that an author be able to agree to be accountable for all aspects of the work, including its accuracy and integrity, and a non-human tool cannot give consent, cannot be held accountable in any meaningful sense, and cannot be contacted about concerns raised after publication1. The Committee on Publication Ethics reached the same conclusion through its own dedicated position statement on authorship and AI tools, emphasizing that accountability, not the quality or usefulness of a given contribution, is the specific test an AI tool fails, since only a person or entity capable of taking responsibility can be an author in any system that relies on authors to stand behind their work2. The World Association of Medical Editors' recommendations on chatbots and generative AI in scholarly publications arrived at the identical core position independently, framing AI tools as something that can be used in the research or writing process but that remains categorically distinct from a contributor capable of being named an author3. This convergence across three separately governed bodies, writing for different disciplinary audiences, is the strongest evidence that the “not an author” conclusion follows from the fundamental purpose of authorship, namely accountability, rather than from the institutional preference of any one organization.
Content of disclosure requirements
Once the authorship question had been settled, editorial policy attention shifted to the more difficult and specific question of what disclosure requires in practice. ICMJE's updated recommendations specify that authors who use AI technology in writing assistance, figure or image generation, or data analysis must report this use in the manuscript, describing which tool was used and how, in enough detail that a reader understands exactly where the technology was involved rather than receiving a single generic acknowledgment covering the whole paper1. WAME's recommendations go further in specifying that authors should be prepared to explain and defend any AI-assisted content exactly as they would any other part of the manuscript, since responsibility for accuracy rests entirely with the human author regardless of which tool helped generate a given sentence, table, or analysis3. Publisher-level policies from Elsevier and Springer Nature implement this same disclosure logic operationally, typically requiring a dedicated statement describing AI tool use to appear in the methods or acknowledgments section rather than being left to author discretion about whether disclosure is even necessary89. In practice, a statement such as “I used ChatGPT to help draft some paragraphs” does not by itself constitute sufficient disclosure under most current policies; the expectation is specificity about which tool was used, for which part of the work, and what checking or revision, if any, the human authors then performed on the output.
Remaining divergence between policies and its significance
Despite the marked convergence on authorship and the general principle of disclosure, specific implementation details still differ across journals in ways that authors should verify individually rather than assume to be standardized. Policies vary on whether AI-generated images or figures are permitted at all, on whether AI-assisted translation of a manuscript originally written in another language needs the same disclosure as AI-assisted drafting, and on exactly where in the manuscript a disclosure statement must appear89. Holden Thorp's January 2023 editorial itself noted that policy would likely need continued revision as the technology and its use cases evolved, a prediction that has been borne out as journals have since issued updated guidance addressing newer capabilities, such as AI-assisted literature searching and AI-assisted peer review drafting, that were not yet common in late 20224. Because a target journal's specific disclosure format and permitted-use boundaries can change between submissions, checking the current version of its instructions for authors at the time of each new submission, rather than relying on a policy read six months or a year earlier, remains the only reliable way to stay compliant with a still-evolving area of editorial policy.
Application of the same principles to AI use in peer review
The same accountability logic that excludes AI tools from authorship has been extended by several of the same editorial bodies to a second, related question: whether reviewers may use AI tools to help draft or generate peer-review reports. COPE's position statement and guidance from individual publishers, including Elsevier, address this specifically, generally prohibiting reviewers from uploading a confidential manuscript into a public or third-party AI tool, since doing so can expose unpublished, confidential research content to a system outside the journal's and authors' control, a distinct confidentiality risk that sits alongside, but is separate from, the authorship and disclosure questions addressed earlier78. This extension illustrates that the underlying principle driving AI policy across the publishing process is broader than authorship alone: wherever a human role in the publication process carries a specific accountability or confidentiality obligation, writing a manuscript, reviewing one, or editing one, current policy asks whether delegating part of that role to an AI tool breaks the accountability or confidentiality chain the role depends on, and answers consistently that it does unless specific safeguards and disclosure are in place.
For reviewers specifically, this has produced guidance that a reviewer may use an AI tool for narrowly limited tasks, such as checking their own draft review for clarity, written using only information the reviewer is personally permitted to share, but should not use a tool to analyze or summarize the confidential manuscript content itself7. Because this distinction between permitted and prohibited AI use for reviewers is more granular than the simpler "disclose if used" rule that applies to authors, and because it continues to be refined as journals gain more experience with how reviewers are actually using these tools in practice, a reviewer invited by any given journal should check that journal's specific, current reviewer guidance on AI use before accepting a review assignment, rather than assuming the author-facing disclosure policy covers the reviewer role as well. Editors themselves face a parallel version of the same question when using AI tools to draft decision letters or summarize reviewer comments, and the same accountability principle has been extended there too: an editor may use an AI tool to assist with routine correspondence, but the editorial judgment itself, and responsibility for it, must remain the editor's own.
Key points
Key points
Every major editorial body converged independently on the same authorship rule: AI tools cannot be authors because they cannot be held accountable. Implementation details concerning disclosure format and permitted uses nevertheless vary by journal and continue to evolve, so the current policy should be checked at each submission rather than assuming that an earlier rule still applies.
Practical application
Practical application
- An AI tool should not be listed as an author or co-author, or in the acknowledgments as a contributor with agency. ICMJE, COPE and WAME all exclude this on the same grounds of accountability; AI use should instead be disclosed as a described method.
- The disclosure statement should be specific rather than generic. The statement should name the tool, describe the purpose for which it was used (drafting, editing, analysis, translation) and note the human checking or revision that followed.
- The current AI policy of the target journal should be checked at the time of each submission. Disclosure format and permitted-use boundaries differ across journals and continue to be updated, so a policy read months ago may no longer reflect the journal's current requirements.
Frequently asked questions
ICMJE's guidance addresses this directly: AI tools should not be listed in the author or acknowledgments section as if they were a contributor with agency; their use should instead be described in the manuscript's methods or a dedicated disclosure statement.
Policies vary on this distinction, and some journals treat routine language-polishing differently from substantive content generation; checking the specific target journal's current instructions for authors is necessary since this is one of the areas where policies have not fully converged.
Under ICMJE, COPE, and WAME guidance alike, the human authors remain fully responsible for the accuracy and integrity of all content in the manuscript, regardless of which tool assisted in generating any part of it.
At least one early instance of ChatGPT being listed as a co-author on a published paper is documented in reporting from this period, which is part of what prompted the rapid, coordinated policy response from editorial bodies in the months that followed.
Yes; editorial commentary from this period, including Science's own January 2023 editorial, anticipated continued revision as AI capabilities and use cases evolved, and journals have since updated guidance to address newer AI applications in research and writing.
References
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 Evolution of AI Authorship Policies in Academic Publishing. Pubrica. https://pubrica.com/academy/citation-and-formatting/ai-authorship-policy-in-academic-publishing/
Pubrica Academy. “The Evolution of AI Authorship Policies in Academic Publishing.” Pubrica, 10 Oct. 2026, pubrica.com/academy/citation-and-formatting/ai-authorship-policy-in-academic-publishing/.
Pubrica Academy (2026) The Evolution of AI Authorship Policies in Academic Publishing. Pubrica. Available at: https://pubrica.com/academy/citation-and-formatting/ai-authorship-policy-in-academic-publishing/ (Accessed: 15 October 2026).
Pubrica Academy. “The Evolution of AI Authorship Policies in Academic Publishing.” Pubrica. October 10, 2026. https://pubrica.com/academy/citation-and-formatting/ai-authorship-policy-in-academic-publishing/
Pubrica Academy, ‘The Evolution of AI Authorship Policies in Academic Publishing’ (Pubrica, 10 October 2026) <https://pubrica.com/academy/citation-and-formatting/ai-authorship-policy-in-academic-publishing/> accessed 15 October 2026.
Pubrica Academy. The Evolution of AI Authorship Policies in Academic Publishing. [Internet]. Pubrica; 2026 Oct 10 [cited 2026 Oct 15]. Available from: https://pubrica.com/academy/citation-and-formatting/ai-authorship-policy-in-academic-publishing/
Pubrica Academy, “The Evolution of AI Authorship Policies in Academic Publishing,” Pubrica, Oct. 10, 2026. [Online]. Available: https://pubrica.com/academy/citation-and-formatting/ai-authorship-policy-in-academic-publishing/
@misc{/la_ai_authorship_policy_academic_publishing_2026,
author = {{Pubrica Academy}},
title = {The Evolution of AI Authorship Policies in Academic Publishing},
howpublished = {Pubrica},
year = {2026},
month = {oct},
url = {https://pubrica.com/academy/citation-and-formatting/ai-authorship-policy-in-academic-publishing/},
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 Template
This template may be adapted for the methods or acknowledgments section of a manuscript.
[Tool name and version] was used to [specific task: e.g., draft an initial outline of the Introduction / check grammar in the Discussion / generate Figure X]. All AI-assisted content was reviewed, verified for accuracy, and edited by the human authors, who take full responsibility for the final manuscript. No AI tool is listed as an author or contributor, consistent with [target journal]'s authorship policy. [ ] Confirmed no AI tool is named as author/co-author anywhere in the manuscript [ ] Named the specific tool and version used [ ] Specified exactly which task(s) it assisted with [ ] Confirmed the target journal's current disclosure format and placement requirements [ ] Documented human review/verification of all AI-assisted content
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 publication ethics team assists authors in drafting AI-use disclosure statements that meet the current, specific requirements of each target journal, reducing the risk that a submission is delayed by a policy mismatch identified after review.