AI Disclosure Statement for Research Papers: When, Why and How to Write One

Most major publishers and the ICMJE now expect authors to disclose AI use, but the rules on when, where, and how differ. This article explains when disclosure is needed, where to place the statement, what to include, how to write one with examples, and how to avoid fake AI-generated references.

Updated on November 19, 2026

AI Disclosure Statement

An AI disclosure statement is a short passage in a manuscript that tells readers, editors, and reviewers which artificial intelligence tools were used in preparing the work, what they were used for, and how the authors checked the results. As generative AI in academic writing has become common, most major publishers and the International Committee of Medical Journal Editors (ICMJE) now expect authors to disclose AI use, although the rules on when to disclose, where to place the statement, and what it must say differ between journals. Understanding those differences before submission helps avoid delays, queries from editors, and, in the worst cases, rejection.

This article covers the following:

  • What an AI disclosure statement is and why journals require one
  • When AI disclosure in research papers is needed, and what usually does not need to be declared
  • Where publishers and the ICMJE ask authors to place the statement
  • What to include, with examples of how to write an AI disclosure statement
  • How to handle AI-generated references, and what to check before citing anything
  • Common mistakes that lead to problems at submission

What Is an AI Disclosure Statement?

An AI disclosure statement is a transparency note. It names the tool or service used, explains the purpose, and confirms that the authors reviewed the output and took responsibility for the final work. It is not a confession or a disclaimer. Used within a journal’s rules, AI tools are accepted by many publishers, and the statement simply records how they were used.

The reasoning behind disclosure is accountability. The ICMJE states that AI tools such as chatbots cannot be listed as authors, because they cannot be responsible for the accuracy, integrity, and originality of a work, and that humans remain responsible for anything submitted that involves AI-assisted technologies. Journals also know that AI output can be incorrect, incomplete, or biased, which makes it important for editors and readers to know where it was used. AJE’s article Best Practices for Generative AI in Research discusses the wider ethics of using these tools in research and writing.

When Is AI Disclosure Needed?

There is no single universal rule, but the major policies point in the same general direction.

Disclosure is generally expected when generative AI is used to:

  • Draft, generate, or substantially rewrite text
  • Summarize or search the literature
  • Write code or analyze data
  • Create, alter, or enhance figures and images

Disclosure is generally not expected for basic language tools. Elsevier states that basic checks of grammar, spelling, and punctuation do not require a declaration. Springer Nature says that “AI assisted copy editing,” which it defines as AI-assisted improvements to human-written text for readability, style, and the correction of errors in grammar, spelling, punctuation, and tone, does not need to be declared, but that generative editorial work and content creation do. Wiley similarly places tools that improve spelling, grammar, and general editing outside the scope of its disclosure requirement.

These exemptions are not universal, and the line between “copy editing” and “generative rewriting” can be blurry. When a tool has done more than correct errors, the safer course is to disclose it. The journal’s current instructions for authors always take precedence over any summary, including this one.

Where to Place an AI Disclosure Statement

Placement differs by publisher. The table below summarizes how several major bodies and publishers ask authors to disclose, based on their published policies at the time of writing. Because these policies are updated frequently, they should be checked again on the journal’s website before submission.

Where to Place an AI Disclosure Statement

The ICMJE recommends that the same information appear in the cover letter, which is why the cover letter is a common place to mention AI use. How to structure that letter is covered in AJE’s guidance on writing a cover letter for journal submission.

What to Include in an AI Disclosure Statement

A useful statement is specific. A line such as “AI was used in preparing this paper” tells an editor very little. Most policies ask for the following information:

  • The name of the tool or service, and, where requested, its version and developer
  • The purpose, meaning what the tool was used for, such as improving the readability of a section, summarizing literature, or classifying data
  • Where it was used, such as which sections of the manuscript or which analysis step
  • How the output was checked, confirming that the authors reviewed, verified, and edited it
  • A statement of responsibility, confirming that the authors take full responsibility for the content

Some journals and bodies go further. The World Association of Medical Editors, for example, encourages authors to detail the prompts used and the version and type of tool, so it is sensible to keep that information on hand.

How to Write an AI Disclosure Statement: Examples

The wording below is illustrative. Bracketed text should be replaced with the actual tool, purpose, and details, and the journal’s required format should be followed where one is given.

Declaration for writing assistance (based on Elsevier’s suggested wording):

During the preparation of this work the author(s) used [name of tool or service] in order to [reason]. After using this tool or service, the author(s) reviewed and edited the content as needed and take(s) full responsibility for the content of the publication.

Statement for use in the methods (data analysis or processing):

[Name of tool, version, developer] was used to [task, such as classify open-ended survey responses]. The prompts and settings used are provided in [supplementary material]. The authors checked the outputs against [verification method] and take responsibility for the results reported.

Statement for the acknowledgments (writing assistance, following ICMJE examples):

The authors used [name of tool] to improve the clarity of language in [section]. All suggestions were reviewed by the authors, who take full responsibility for the final text.

Sentence for the cover letter:

The use of [name of tool] for [purpose] is disclosed in the [Acknowledgments or Methods] section of the manuscript.

AI-Generated References and Fake Citations

One of the most serious risks of using generative AI in research writing is the invention of references. ChatGPT and other tools can produce fake references that look entirely convincing, with plausible authors, titles, journals, and even DOIs, but that do not correspond to any real publication. Elsevier’s guidance for authors warns that AI tools can invent citations that look real but do not exist.

AI-generated references should therefore never be trusted without verification. Before any source is cited:

  • Look the source up in a database or on the publisher’s website, not in the AI tool itself
  • Confirm that the DOI resolves to the paper that is cited
  • Check that the authors, title, journal, year, volume, and pages all match
  • Read the source, and confirm that it actually supports the claim it is cited for

A manuscript with fabricated references can be rejected, and the author remains responsible for every citation regardless of how it was produced. AJE’s 2023 Guide to Avoiding Plagiarism: How to Avoid Plagiarism in the World of AI covers related risks.

Can I Use ChatGPT to Write a Paper?

Whether a researcher can use ChatGPT to write a paper depends on the journal’s policy, and the answer is not a simple yes or no. Many publishers permit the use of generative AI tools if their use is disclosed and the authors remain fully responsible for the content. All of the policies above agree that AI tools cannot be listed as authors. Some publishers restrict particular uses, such as generating images, and some journals and funders have stricter rules than the publisher’s general policy.

Confidentiality is a separate concern. Entering unpublished data or manuscript text into a public AI tool may expose it, and Elsevier advises authors to check a tool’s terms before uploading anything. Authors who intend to use AI at any point should first read the journal’s policy and not after the manuscript is written.

Keep a Record as You Write

The easiest way to write an accurate disclosure is to record AI use as it happens. A simple log that notes the tool, version, date, what it was used for, and which part of the work it affected makes the final statement quick to write and easy to defend if an editor asks follow-up questions. It also helps when a manuscript moves to a different journal, since the new journal may ask for different details.

Common Mistakes to Avoid

  • Assuming a grammar checker exemption also covers generative rewriting or drafting
  • Writing a vague AI disclosure statement that does not name the tool or its purpose
  • Placing the statement in the wrong section, or disclosing in the manuscript but not in the cover letter when the journal asks for both
  • Citing AI-generated references without verifying that they exist and support the claim
  • Listing an AI tool as an author or co-author
  • Uploading unpublished data or manuscript text to a tool without checking its terms
  • Failing to recheck the journal’s policy when resubmitting to a different journal, since requirements differ and change

Final Thoughts

An AI disclosure statement is a small piece of text with a clear job: to show editors and readers exactly how AI tools contributed to a piece of research, and that the authors own the result. The rules differ across publishers, which makes it important to read the target journal’s policy, record AI use as it happens, and verify everything an AI tool produces. Researchers who want a manuscript to be clear and well written before submission can also use AJE’s english editing services, while the declaration of AI use remains the author’s responsibility.

Contributors
Tag
Table of contents
Share+
FacebookTwitterLinkedInCopy linkEmail
Join the newsletter
Sign up for early access to AJE Scholar articles, discounts on AJE services, and more

See our "Privacy Policy"