AI Use Disclosure has become a practical trust issue for authors, editors, reviewers, and publishers. The recent Oxford University Press survey, published on September 11, 2026 and covering about 2,600 researchers globally, found that 74% of respondents were unclear about what AI use they should disclose, according to the OUP survey report. That finding points to a problem authors can address before submission: do not wait for perfect industry agreement before keeping records and making clear, limited, factual statements about AI assistance.
Why AI Use Disclosure Feels Unclear To Authors
AI Use Disclosure Starts With Scope
Authors are not all using AI tools in the same way. Some use them for language polishing. Some use them to summarize notes, test outlines, translate passages, prepare plain-language descriptions, compare titles, or draft promotional copy. Others may use AI in ways that touch analysis, interpretation, figures, or manuscript drafting. The disclosure question changes depending on where the tool entered the work.
The OUP finding on uncertainty matters because a vague policy can leave careful authors guessing. If a publisher, journal, funder, or conference asks only whether AI was used, the answer may be too blunt to be useful. A stronger process asks what tool was used, where it was used, how much it shaped the submitted work, and who checked the output.
Fear Can Reduce Transparency
The same OUP-related reporting said 44% of respondents feared that disclosing AI use would negatively affect how their work was judged, while only 36% kept any record of how they used AI during research, as reported by Library Learning Space on the OUP study. Those two findings belong together. If authors believe disclosure will be punished, and if they have no records, they may delay disclosure until the submission form forces a rushed answer.
That is a weak position for any author. Disclosure should not be framed as a confession of poor work. It is better understood as a factual note about tools used in the production process. The author still remains responsible for accuracy, originality, permissions, citation quality, and the final wording submitted under their name.
What Authors Should Record Before Submission
A Practical Record For Each Project
The strongest AI Use Disclosure begins before the manuscript is finished. Authors do not need a burdensome diary, but they do need enough information to answer reasonable editorial questions later. A simple project log can prevent memory gaps and reduce the risk of inconsistent statements across a proposal, manuscript, peer review response, and marketing package.
- Tool name and version if available, with the date range of use.
- Stage of work affected, such as outlining, language editing, translation, summarizing, image preparation, coding support, or promotional copy.
- Specific sections affected, such as abstract, introduction, figure caption, bibliography check, author biography, or press materials.
- Type of input provided to the tool, avoiding confidential material if a publisher or institution prohibits that use.
- Human review steps, including fact checking, citation checking, revision, and removal of unsupported claims.
- Final disclosure wording used in submission forms, acknowledgments, cover letters, or production notes.
This record is useful even when a publisher later decides that a minor use does not require a formal note. The value is readiness: the author can respond clearly, rather than trying to reconstruct tool use after acceptance or during production.
Meaningful Use Needs Clear Boundaries
Not every use carries the same weight. An author who uses a spelling aid is in a different position from an author who asks a generative tool to draft a discussion section or interpret data. Still, the safest editorial habit is to define categories in advance: minor language support, structural support, research administration, content generation, data-related assistance, visual or figure support, and promotional support.
That categorization helps authors avoid two common errors. The first is over-disclosing in vague language that sounds alarming but says little. The second is under-disclosing because the author assumes a tool was only used in a background step. A useful AI Use Disclosure statement is specific enough to explain the use without overstating the tool’s contribution.
How To Write A Clear Disclosure Statement
Keep The Statement Factual And Limited
A disclosure statement should be plain. It should not market the tool, apologize for using it, or imply that the tool is responsible for scholarly judgment. A practical version might read: The author used a generative AI tool during language editing of the introduction and conclusion. The author reviewed and revised all AI-assisted text and takes responsibility for the final manuscript.
For a book proposal, an author might adapt the same structure: The author used AI assistance to organize chapter summaries and test alternative subtitle wording. No AI-generated text was submitted without human review. For a journal article, the statement may need more detail if the tool touched analysis, figures, translations, or submitted text. Authors should follow the exact policy of the journal, publisher, funder, or institution receiving the work.
Match The Statement To The Submission System
One weakness identified in the research notes is that disclosure is not always requested at the point of submission. The OUP survey reporting included the finding that only 42% of journal authors and 30% of book authors said they were asked about AI use. If the form does not ask, authors can still keep the record and raise the point in a cover letter, author questionnaire, production note, or editor correspondence when the use is material.
Collaborative projects need even more care because co-authors may use different tools at different stages. A shared log, a named person responsible for collecting entries, and a pre-submission check can prevent conflicting answers. For teams working on manuscripts together, related guidance on team disclosure requirements can help turn individual notes into a common process.
Policy Questions Authors Should Ask Publishers
Submission Forms Should Not Be The First Conversation
Authors can ask practical questions before a manuscript is ready: Which AI uses must be declared? Are language editing, translation, summaries, figures, data analysis, and promotional materials treated differently? Where should the disclosure appear? Does the publisher require a statement even if AI was used only outside the manuscript, such as for marketing copy or an author biography?
These questions are not adversarial. They help publishers, authors, and editors work from the same expectations. For authors comparing resources on publishing practice and author support, Leap Year Publishing offers insights as a related site in the same network that may be useful alongside publisher-specific policies.
Discipline Norms Still Need Care
The research notes show that disclosure habits vary across fields and formats. They also indicate gaps between general AI use and formal disclosure rates in manuscripts. That gap does not prove bad faith in every case. It may reflect unclear rules, inconsistent forms, fear of stigma, or uncertainty about whether minor uses count.
For cross-disciplinary authors, this creates a real publishing challenge. A historian, medical researcher, education scholar, and trade nonfiction author may face different expectations even when using similar tools for language review. Authors should not assume that a practice accepted in one field will be accepted in another. The safer method is to keep records, read the current policy for each venue, and ask concise questions before submission.
AI Use Disclosure For Author Trust
AI Use Disclosure is not only a compliance task. It is part of the relationship among authors, editors, reviewers, readers, and institutions. Clear disclosure tells others where assistance entered the work and where human responsibility remained. That distinction is especially important in scholarly and professional publishing, where trust depends on traceable claims, accurate citations, and honest representation of the writing process.
Authors do not need to predict every future policy change to act responsibly now. They can keep a project log, separate minor language help from substantive drafting or analysis, review all output carefully, and use direct wording in the places where publishers request it. If a policy is unclear, authors can ask for clarification and save the response with their project files.
Treat AI Use Disclosure as part of publication preparation, not as a last-minute box on a submission form. The survey findings show that uncertainty, fear, and weak record-keeping are common. Authors can reduce all three by building a steady disclosure habit before the manuscript reaches an editor.