AI Ethics Frameworks for Academic Publishing: A Complete Guide to Disclosure, Accountability & Authorship Rules

Artificial intelligence is increasingly used to draft, edit, summarize, and refine academic manuscripts. While these tools improve efficiency, they also raise concerns about authorship, transparency, and research integrity. To address these issues, professional organizations and leading academic publishers have introduced AI governance frameworks that establish clear rules for disclosure, accountability, and responsible use.

The following frameworks outline how researchers can ethically incorporate AI into scholarly publishing while ensuring that human authors remain responsible for their work.

ACM Code of Ethics (2018)

The Association for Computing Machinery (ACM) Code of Ethics and Professional Conduct provides ethical principles for computing professionals, including AI researchers and developers. Although published before the rise of generative AI, its guidance remains highly relevant to AI research.

Core Principles

The Code emphasizes avoiding harm, acting honestly, respecting privacy, and maintaining professional competence. These principles require AI developers to evaluate systems for bias, security risks, and unintended consequences before deployment. Researchers are also expected to disclose limitations and avoid presenting misleading or fabricated AI-generated information.

Applying the Code in Practice

For example, a team developing an AI diagnostic tool should test the model across diverse datasets, report known limitations, and ensure the technology does not unfairly disadvantage particular groups. By promoting accountability and transparency, the ACM Code continues to provide a strong ethical foundation for responsible AI development.

Committee on Publication Ethics (COPE) AI Guidance (2023)

The Committee on Publication Ethics (COPE) published AI guidance in 2023 to help journals, editors, reviewers, and authors manage the responsible use of generative AI in scholarly publishing.

Authorship and Disclosure Rules

COPE states that AI tools cannot be listed as authors because they cannot take responsibility for published research. Human authors remain fully accountable for the accuracy, originality, and integrity of their manuscripts. The guidance also recommends disclosing any significant use of AI during manuscript preparation so editors and readers understand how the technology contributed.

Protecting Confidential Manuscripts

COPE further advises reviewers and editors not to upload confidential manuscripts into public AI systems because doing so may compromise confidentiality and intellectual property.

For example, an author who uses AI to improve grammar or summarize sections of a manuscript should disclose this assistance while verifying that every fact, citation, and conclusion remains accurate.

International Committee of Medical Journal Editors (ICMJE) Recommendations (2023)

The International Committee of Medical Journal Editors (ICMJE) updated its recommendations in 2023 to address generative AI in medical publishing. Because medical research directly affects patient care, the committee places strong emphasis on transparency and accountability.

Why AI Cannot Be an Author

ICMJE specifies that AI cannot qualify as an author since it cannot meet authorship requirements or accept responsibility for published work. Researchers must disclose AI use during manuscript preparation and clearly describe how the technology was used.

Verifying AI-Generated Medical Content

The committee also warns authors not to rely on AI-generated medical information or references without independent verification because language models may produce inaccurate or fabricated content.

For instance, authors may use AI to improve the readability of a medical review article, but they remain responsible for confirming every citation, statistical result, and clinical recommendation before publication.

Elsevier Generative AI Author Policy (2023)

Elsevier introduced its Generative AI Author Policy in 2023 to clarify how AI may be used during manuscript preparation. The policy permits AI to improve grammar, language, and readability but prohibits using AI as a substitute for scientific analysis or human judgment.

Authorship and Disclosure Requirements

Elsevier states that AI cannot be credited as an author because it cannot assume ethical or legal responsibility for research. Authors must disclose significant AI assistance where required and carefully review all AI-generated content before submission.

Verification of References

The publisher also cautions researchers against relying on AI-generated references or factual claims without verification since generative AI can produce fabricated information.

For example, an author may use AI to edit the language of a research paper but must verify every citation and remain fully responsible for the manuscript’s content.

Springer Nature AI Policy (2023)

Springer Nature established one of the earliest publisher policies governing generative AI in scholarly publishing. The policy allows AI to assist with language editing but requires transparency whenever AI significantly contributes to manuscript preparation.

Accountability for Human Authors

The publisher clearly states that AI tools cannot be named as authors because they cannot accept responsibility for research. Human authors remain accountable for the originality, accuracy, and integrity of every submission.

Confidentiality for Peer Reviewers

Springer Nature also prohibits peer reviewers from uploading confidential manuscripts into generative AI systems to protect unpublished research and intellectual property.

For example, researchers submitting to a Springer Nature journal may use AI to improve writing quality, but they must disclose its use where appropriate, verify all AI-generated content, and ensure the final manuscript reflects their own scientific judgment.

Wiley AI Guidelines (2023)

Wiley introduced its AI Guidelines in 2023 to help authors, editors, and peer reviewers use generative AI responsibly while protecting research integrity. The guidelines acknowledge that AI can improve writing quality and productivity but stress that it should never replace human expertise or accountability.

Authorship Standards

Wiley prohibits listing AI systems as authors because they cannot take responsibility for published research or meet accepted authorship standards. Authors remain fully accountable for the originality, accuracy, and integrity of every part of a manuscript, including content generated or edited with AI. The publisher also expects researchers to disclose significant AI use during manuscript preparation, allowing editors and readers to understand the technology’s role.

Guidance for Editors and Peer Reviewers

For editors and peer reviewers, Wiley advises against entering confidential manuscripts into public AI tools without authorization, as doing so could compromise sensitive research and intellectual property.

For example, a researcher may use AI to improve grammar and readability before submitting an article to a Wiley journal. However, the author must verify all facts, citations, and conclusions and disclose AI assistance where required.

Taylor & Francis AI Policy (2023)

The Taylor & Francis AI Policy provides guidance for the ethical use of generative AI in scholarly publishing. The policy aims to balance the benefits of AI-assisted writing with the need to maintain transparency, originality, and author responsibility.

Disclosure and Authorship

The publisher allows authors to use AI for language editing and writing support but requires that its use be disclosed whenever it has made a meaningful contribution to manuscript preparation. AI tools cannot be credited as authors because they cannot accept responsibility for the published work or comply with established authorship requirements.

Reviewing AI-Assisted Content

Taylor & Francis also emphasizes that authors must carefully review AI-generated content to ensure that references, data, and scientific interpretations are accurate. Editors and reviewers are advised not to upload confidential manuscripts into AI platforms unless permitted by journal policies.

For instance, an author using AI to refine the language of a journal article should disclose this assistance and ensure that the final manuscript reflects their own analysis and conclusions.

Common Themes Across the Frameworks

ACM Code of Ethics, COPE, ICMJE, Elsevier, Springer Nature, Wiley, and Taylor & Francis share several fundamental principles for the ethical use of AI in academic publishing:

• AI cannot be recognized as an author because it cannot assume legal, ethical, or intellectual responsibility for published research.
• Human authors remain fully accountable for the accuracy, originality, and integrity of all manuscript content, including AI-assisted writing.
• Transparency is essential, with most frameworks requiring or recommending the disclosure of significant AI use during manuscript preparation.
• AI-generated content must be verified to ensure that facts, references, data, and conclusions are accurate and free from fabricated information.
• Confidentiality must be protected, meaning editors and peer reviewers should not upload unpublished manuscripts to public AI platforms without authorization.
• AI should support—not replace—human judgment, serving as a tool for tasks such as language editing and content refinement rather than independent research or decision-making.
• Research integrity remains the primary objective, ensuring that AI is used responsibly while maintaining trust, credibility, and ethical standards in scholarly publishing.

Key Takeaways

Although these AI governance frameworks were developed by different professional organizations and publishers, they share several consistent principles. First, none of the frameworks recognize artificial intelligence as an author because AI cannot assume legal or ethical responsibility for scholarly work. Second, they require transparency through appropriate disclosure whenever AI significantly contributes to manuscript preparation. Third, all frameworks emphasize that human authors remain fully accountable for verifying AI-generated content, including facts, references, and conclusions. Finally, editors and peer reviewers are encouraged to protect confidential research by avoiding the use of public AI systems when handling unpublished manuscripts.

Together, these principles help preserve research integrity while enabling researchers to benefit from generative AI in a responsible and ethical manner.

Conclusion

Generative AI is transforming academic publishing, but its use must be guided by clear ethical standards. The ACM Code of Ethics, COPE, ICMJE, Elsevier, Springer Nature, Wiley, and Taylor & Francis all emphasize the same fundamental principle: AI is a tool that can support research and writing, but it cannot replace human responsibility. By requiring disclosure, protecting confidentiality, and maintaining author accountability, these frameworks promote transparency and trust in scholarly communication. As AI technologies continue to evolve, following these established guidelines will remain essential for preserving the credibility and integrity of academic research.