Generative AI Policy – Y Academy Press

About this policy

Y Academy Press writes, edits, produces, and publishes within one house. Rules that separate editorial, review, and production roles at a large publisher have to be restated here as rules about which activity is being performed, not about who is performing it. That is the shape of this policy.

It is modelled on the generative AI policies published by Elsevier for its journals and for its books and commissioned content, adapted to a small scholarly imprint that also teaches.

What it covers

Category	Work covered
Press books	Trade nonfiction, textbooks, Editorial editions, and commissioned content
Scholarly papers and theses	AOM-format papers, dissertations, and conference papers prepared under either mark
Academy teaching and research	Course materials, workshop and certificate material, internal research
Assessed student work	Coursework, assignments, and examinations submitted to the Academy

It binds authors, co-authors, and contributors; translators; volume and commissioning editors; illustrators and designers; peer reviewers and editorial readers; Press editorial and production staff; and Academy faculty and students.

Precedence

Where a work is submitted to a third party, that body's AI policy governs and this one is the floor, not the ceiling. The Academy of Management, university graduate offices, and external journals each publish their own rules and revise them without notice. Confirm the current version of the receiving body's policy before submission. Do not rely on a summary, including this one.

The four governing principles

Everything else in this policy follows from these, and any case the policy does not anticipate should be decided by reference to them.

Responsibility does not transfer. An AI system is an instrument. Accountability for every claim, figure, citation, and image rests with the person who submits the work. This is the reason an AI tool cannot hold authorship: authorship carries obligations, approval of the final version, agreement to submission, answerability for error, that only a person can discharge.
Disclosure is proportionate to substantive contribution. Checking spelling does not require a declaration. Restructuring an argument does. The test is not how long a tool was used or how many words passed through it, but whether its output changed what the work says.
Confidentiality is not waived for convenience. An unpublished manuscript belongs to its author. Placing it into a tool that retains or trains on its inputs discloses someone else's property to a party who was never granted it, and where the material carries personal information, may breach privacy law. The obligation is not lifted because a tool is faster or already open in a browser tab.
Verification precedes publication. Generative systems produce fluent text that is sometimes false and references that are sometimes invented. Every checkable item must be confirmed against a primary source before it enters a Press text.

Definitions

Term	Meaning in this policy
Generative AI	Technology that produces content, including text, images, audio, code, and synthetic data, in response to a prompt. Examples current at the date of this version include ChatGPT, Claude, Gemini, Microsoft Copilot, and DALL·E.
AI-assisted technology	A tool that applies AI to an artifact rather than generating new content: grammar and spelling checkers, translation engines, reference managers, transcription, and assistive-access software.
Private AI tool	A deployment that does not retain, re-use, share, or learn from submitted content beyond what is needed to deliver the service, does not use inputs to train or improve a model, and does not pass them to a third party.
Explanatory image	Illustrates a concept, process, or relationship: flow charts, decision trees, timelines, schematic illustrations, workflow diagrams.
Data visualization	Presents results derived from a dataset through an analytical, computational, or statistical workflow: plots, charts, graphs, heatmaps.
Primary research or documentary image	Represents observed or experimental data, or a real scene: micrographs, scans, photographs, archival reproductions.
Declaration statement	The tool-and-purpose statement required of authors and reviewers.
Gradient disclosure	The Y Academy house statement of the character of an AI collaboration.

1For authors and contributors

Permitted uses

Synthesizing literature, gaining research insight, and identifying gaps in a field.
Generating, testing, and pressure-testing ideas the author then judges.
Organizing or restructuring material the author has written.
Improving language, clarity, and readability.
Checking a manuscript for completeness and against submission guidelines.
Translation, subject to the tool-terms rule below and to verification of the translated text by a competent reader.
Assisting with certain images, subject to section 2.

Required oversight

Review all output for accuracy, completeness, and bias before it enters the manuscript.
Confirm that every reference exists, is correctly attributed, and says what it is cited for. Fabricated references are the single most common defect in AI-assisted scholarly drafts.
Ensure the text reflects the author's own analysis, interpretation, and insight, not an assembled approximation of it.
Edit the material thoroughly enough that the work is an authentic and original contribution.

Prohibited

Fabricating or altering data, results, or references.
Presenting AI output as the author's analysis where the author has exercised no judgement over it.
Generating whole sections in which the author has made no genuine intellectual contribution.
Uploading third-party copyrighted material, including a previous edition under revision, where the tool's terms do not permit it.
Uploading personal or identifiable information about research participants or third parties.

Tool terms and rights

Before using any tool on unpublished material, confirm from its terms and conditions that the confidentiality of the material is maintained; that the tool takes no rights beyond those needed to deliver the service; that nothing in the terms restricts subsequent publication; and that inputs are not used to train or improve a model. Where these cannot be established, use a private tool as defined in section 0.

Authorship and citation

AI tools are not listed as authors or co-authors, are not named in an author line, and are not cited as the author of a source. Where AI output is quoted or reproduced in the text, it is identified as such, and the tool, version, date, and prompt are recorded.

The declaration

Where AI use exceeded the threshold set by the second governing principle, the author submits a declaration titled Declaration of Generative AI and AI-assisted Technologies, recording the tool name and version, the purpose of use, and the extent of the author's oversight. The declaration appears in the published work. It is not an internal note.

Declaration template

During the preparation of this work the author(s) used [tool name and version] in order to [purpose]. After using this tool the author(s) reviewed and edited the content as required and take(s) full responsibility for the content of the published work.

Where the declaration goes

Format	Placement
Trade book	Author's Note in the front matter, with the disclosure line on the copyright page
Textbook	Preface or front-matter note, with the disclosure line on the copyright page
Scholarly thesis	Front matter, immediately following the title page
AOM or journal paper	Title-page insert and acknowledgements, and in the cover letter where the journal requires it
Report or course material	End of the document, immediately above the references

What needs no declaration

Basic checks of grammar, spelling, and punctuation.
AI features within assistive technology used for accessibility.

Where AI forms part of the research method rather than the writing, it is not declared here at all. It is described in the methods, with the model or tool name, the version, and the developer or manufacturer, in enough detail to be reproduced.

The house gradient disclosure

Y Academy's own disclosure practice predates this policy and is retained by it. Where a book, thesis, or paper is written under either mark with sustained AI assistance, the front matter carries the gradient disclosure. It and the declaration do different work and are not substitutes: the gradient statement describes the character of the collaboration for a reader, while the declaration records tool, purpose, and oversight for a publisher, examiner, or institution. Where a third party requires the declaration form, both appear.

Gradient disclosure · personal attribution

The framework and argumentative spine of this book are the author's own. The text as produced reflects a gradient of contribution developed in sustained dialogue with an AI system used as an instrument of drafting and pressure-testing.

Responsibility is asymmetric: the judgements, the claims advanced, and the errors that remain are the author's, while the AI's role was to draft, challenge, and refine language under the author's direction. This note discloses that collaboration in the belief that transparency strengthens rather than weakens the work.

2Images and artwork

Three categories, three different rules. The category is fixed by what the image represents, not by how it was made or where it appears.

AI may assist

Explanatory images

Flow charts, decision trees, timelines, schematic conceptual illustrations, workflow diagrams.

Disclose the tool, its version, and how it was used in the image caption, and record the use in the declaration. The author remains responsible for the accuracy of what the diagram asserts.

Conditional

Data visualizations

Plots, charts, graphs, heatmaps.

Permitted only where the visual output is directly derived from the underlying data through reproducible analytical, computational, or statistical methods. Never to fabricate results, invent or alter data, or generate a figure not faithfully derived from the data. Disclose model, version, and developer in the methods or a figure note.

Not permitted

Primary and documentary images

Micrographs, scans, photographs, archival reproductions.

Generative AI must not create or alter an image representing observed or experimental data, or a real scene. No enhancing, obscuring, moving, removing, or introducing a feature.

Adjustment and the research exception

Adjustments to brightness, contrast, or colour balance are acceptable only through established image-processing software, and only where no information present in the original is obscured or eliminated.

None of this prevents the use of AI within a formal research design or method, for example an AI-assisted imaging approach used to collect or interpret the underlying data. Where AI is part of the method, it is described reproducibly in the methods, with model name, version, and developer or manufacturer. Authors may be asked to supply original, unprocessed files, and the Press may apply image forensics to submitted artwork.

Covers, graphical abstracts, and brand artwork

General-purpose generative AI image tools are not used to create graphical abstracts. Use dedicated scientific illustration tools, whose licensing terms are defined and which support responsible use of copyrighted material.
Cover art in which a generative AI tool played any part requires prior written permission from the Press before submission. The author must demonstrate that all necessary permissions for third-party material have been obtained and that appropriate attribution is provided.
Covers and interior artwork produced by the Press itself under the brand standards are Press production and fall under section 5. Where a generative tool contributed to a published cover, the copyright page records it.
The imprint marks, the crest, the motto ribbon, and the wordmarks are never generated, redrawn, traced, or restyled by an AI tool, in any context, by anyone.

Prohibited in every category

No image may be generated that duplicates or derives from an existing copyrighted image without permission; that depicts a real, identifiable person; that reproduces a third party's identifiable products or brands; or that synthesizes any likeness of an individual's voice.

Authors are responsible for the accuracy and originality of every image submitted, for confirming that the concepts an image presents are correct, and for obtaining attribution and rights clearance before submission rather than after acceptance.

3For reviewers

A manuscript sent for review is a confidential document. Nothing in it may be disclosed, quoted, or transmitted outside the review.
Do not upload a submitted manuscript, any part of it, or a review questionnaire, report, or correspondence containing manuscript content into an AI tool that is not private. Doing so may violate the author's confidentiality and proprietary rights and, where the material carries personal information, may breach data privacy law.
AI cannot replace critical reading or independent evaluation, and carries a real risk of incorrect, incomplete, or biased conclusions. The reviewer remains fully responsible for the content of the report.
Permitted supportive use, with private tools only: improving the language and structure of a report the reviewer has already written, and assisting with a background literature search. Not permitted: evaluating the manuscript, forming the recommendation, or drafting the substance of the assessment.
Disclose in the report the tool used and the purpose. Basic spelling and grammar checks need no disclosure.
Before using any tool, read its privacy policy and terms for statements on data retention and model training. Prefer an enterprise, paid, or locally hosted deployment. Free consumer tiers are usually not private by default. Check whether an institutional deployment is available first.

Reviewer disclosure template

During the preparation of this report I used [tool name and version] in order to [purpose]. After using this tool I reviewed and edited the content as required and I take full responsibility for its content.

4For editors

A manuscript under evaluation is a confidential document.
Do not upload an author's manuscript, or editorial correspondence about it, into an AI tool that is not private.
AI does not make, and does not substitute for, an editorial decision. The editor remains accountable for the editorial process, for the decision, and for how it is communicated to the author.
Permitted supportive use, with private tools only: improving the language and structure of decision letters, background literature searching, and administrative completeness checks.
Authors are told, in the author guidelines and in the publication agreement, that AI tools may be used in a supportive capacity in the editorial and production process.

The production exception

The Press's production pipeline processes author manuscripts through AI-assisted tools for text extraction, defect repair, house formatting, and typesetting. That is production, not evaluation, and the rule above does not prohibit it. It does, however, mean an author's unpublished manuscript is placed into an AI system, which the third governing principle does not permit to happen silently.

Consent

The author's informed written consent is obtained in the publication agreement before production begins, and the tools used must satisfy the private-tool definition. No title enters production without that consent on record.

5Publication and production

Y Academy Press uses AI-assisted tools across its editorial and production process to improve consistency, accuracy, and turnaround, including in manuscript preparation, formatting, and pre-publication checking. Human oversight remains at the core of every editorial decision. The imprint, not the tool, is accountable for what it publishes.

Tools used on author material are required to meet the confidentiality standard set out in section 0, and their use in any given title is disclosed to the reader through the Author's Note and the copyright-page disclosure line. The extent of AI use varies by title and by format. An author with questions about a specific title should raise them with the Press before signing.

This edition was produced with the assistance of an AI system used as an instrument of drafting and production under Y Academy Press editorial direction; see the Author's Note.

Review and custody

Custodian: Dr. Youssef Ahmad Youssef. This policy is reviewed annually, and on any material change to the policies of the publishers, journals, or institutions to which Y Academy work is submitted.

6Teaching and assessed work

Publication tests a claim. Assessment develops a capacity. A tool that improves a manuscript can, in a course, prevent the formation of the judgement the course exists to build. Permitted use in assessed work is therefore narrower than in publication, and narrower for reasons that have nothing to do with dishonesty.

Coursework. Permitted use is set by the course outline for each course. Where the outline is silent, the declaration requirement applies and students declare the tool, the purpose, and the extent of use. A student unsure whether a use requires declaration should declare it.
Faculty materials. Faculty producing course materials, slides, assessments, and certificates under the Academy mark follow the author rules in sections 0, 1, and 2. Material distributed to students that was substantially AI-drafted carries a disclosure line.
Student work is confidential. Student submissions are not uploaded into a non-private AI tool for marking, feedback, or screening. The confidentiality principle applies to a student's work exactly as it applies to an author's manuscript.

7Breach and correction

Before publication. Undisclosed use discovered before publication is remedied by adding the declaration, and the schedule may slip as a result. Where the undisclosed use is fabrication, the work is not published.
After publication. A disclosure failure is remedied by a correction notice appended to the edition. Fabricated or manipulated data, references, or images are remedied by withdrawal of the edition.

On detection tools

The Press does not treat the output of AI-detection or AI-text-classifier tools as evidence of a breach. Their error rates, in both directions, are too high to carry a finding against an author, and they misclassify the writing of non-native English speakers at higher rates than the writing of native speakers. A finding rests on the record: source documents, drafts, correspondence, and, for images, forensic examination of the original files.

When in doubt, declare. An author, reviewer, editor, or student who is unsure whether a use requires declaration should declare it. Nothing in this policy penalizes over-disclosure, and no one has ever been harmed by having said more than was strictly required about how their work was made.

8Frequently asked questions

For authors

I used a spelling and grammar checker. Do I have to declare it?

No. Basic checks of grammar, spelling, and punctuation need no declaration, and neither does AI inside assistive technology used for accessibility. The threshold is whether the tool's output changed what the work says. A checker that corrects "recieve" did not. A tool that reorganized your chapter did.

How much AI use is too much? Will heavy use get my book rejected?

Extent is not the test, and the Press does not apply a percentage threshold. Two things get a work rejected: fabrication, and the absence of the author's own judgement. A book substantially drafted in dialogue with an AI system, where the framework, the argument, and the decisions are the author's and every claim has been verified, is publishable and is declared. A book assembled from output the author did not evaluate is not, however little of it was generated.

Does the gradient disclosure replace the declaration, or do I need both?

They do different work. The gradient disclosure tells a reader what kind of collaboration produced the text. The declaration tells a publisher, examiner, or institution which tool was used, for what, and with what oversight. For a Press book the gradient disclosure in the Author's Note is the primary instrument. Where a third party requires the declaration form as well, both appear.

I used AI in the research itself, not in the writing. Where does that go?

In the methods, not in the declaration. Describe it reproducibly: model or tool name, version, developer or manufacturer, and the parameters and processing steps someone would need to obtain the same result. This applies to AI-assisted imaging, computational analysis, and any tool that formed part of the research design.

Can I list the AI tool as a co-author, or cite it as a source?

No to both. Authorship carries obligations that only a person can discharge: approving the final version, agreeing to submission, and answering for error. Where you quote or reproduce AI output in the text, identify it as such and record the tool, version, date, and prompt.

Can I use AI to translate my manuscript?

Yes, subject to two conditions. The tool's terms must satisfy the confidentiality and rights requirements in section 1, and a reader competent in both languages must verify the result. Machine translation fails in ways that are invisible to someone who reads only the output, and the author remains responsible for what the translated text says.

Does AI-assisted coding fall under this policy?

Yes, and where it lands depends on what the code does. Code that is part of the research, an analysis script or a model, is described in the methods or the software-availability statement. Code that produced text or figures appearing in the work is covered by the declaration. In both cases the author is responsible for the code being correct, not merely for it running.

I am not sure whether my use crosses the threshold. What should I do?

Declare it. Nothing in this policy penalizes over-disclosure, and a declaration that was not strictly required has never harmed a work. An omitted declaration that was required means, at minimum, a delay, and after publication a correction notice.

Images and artwork

Can I make my diagrams with an AI tool?

Yes, for explanatory images: flow charts, decision trees, timelines, schematic illustrations, workflow diagrams. Disclose the tool, its version, and how you used it in the image caption, and record it in your declaration. You remain responsible for whether the diagram asserts something true.

Can I clean up a photograph or a scan before submitting it?

Only through established image-processing software, and only where nothing present in the original is lost. Brightness, contrast, and colour balance may be adjusted. Enhancing, obscuring, moving, removing, or introducing a feature may not, whatever tool is used. Keep the original unprocessed file; you may be asked for it.

I have generated a cover image I like. Can the Press use it?

Only with prior written permission from the Press, obtained before submission. You will need to demonstrate that all necessary permissions for any third-party material have been obtained and that attribution is provided. Permission is not automatic, and covers the Press produces itself follow a separate route under section 5.

Why the stricter rule for graphical abstracts?

Licensing. Dedicated scientific illustration tools carry defined terms covering the copyrighted and intellectual property their libraries contain. General-purpose image generators do not, which leaves the provenance of what they produce unresolved at exactly the point where a figure stands as the public face of the work.

For reviewers and editors

Why can I not put a manuscript into a chatbot to help me review it?

Two reasons, and the first is not about quality. A manuscript under review is someone else's unpublished property, and uploading it to a tool that retains or trains on its inputs discloses it to a party who was never granted it. Where it carries personal information, that may also breach privacy law. The second reason is that the assessment is the thing you were asked for, and a system that can produce a confident, incorrect reading cannot supply it.

What exactly counts as a private AI tool?

One that does not retain, re-use, share, or learn from what you submit beyond what is needed to deliver the service, does not use your inputs to train or improve a model, and does not pass them to a third party. To check: read the privacy policy and terms for retention and training statements. Assume a free consumer tier is not private by default. Enterprise, paid, and locally hosted deployments are more likely to qualify, and an institutional deployment, if one is available to you, is the first place to look.

So what may I use AI for as a reviewer?

With a private tool: improving the language and structure of a report you have already written, and background literature searching. Not: evaluating the manuscript, forming the recommendation, or drafting the substance of the assessment. Disclose the tool and the purpose in your report.

Teaching and assessment

May students use AI in their coursework?

The course outline governs, and it may permit more or less than this policy does. Where the outline is silent, declare the tool, the purpose, and the extent. The restriction in a course is not a suspicion of dishonesty. It exists because the friction the tool removes is often the thing that was developing the capacity the course is for.

Does the Academy use AI detectors on student work?

No. Detector output is not treated as evidence of undeclared use, and student submissions are not uploaded into non-private tools for marking, feedback, or screening. Where a question arises, it is resolved on the record: drafts, sources, and a conversation with the student.

Showing the single result