# How to disclose ai use in published work?

Brooklyn Bishop · November 7, 2026

> Navigating Vayena's 84 Ethics Guidelines The landscape of artificial intelligence disclosure remains fragmented, with major institutions offering...

## Navigating Vayena's 84 Ethics Guidelines

The landscape of artificial intelligence disclosure remains fragmented, with major institutions offering conflicting advice that leaves many authors uncertain about their obligations. While some medical journals explicitly discourage mandatory disclosure, others enforce strict transparency requirements for academic publications. This inconsistency creates significant anxiety for researchers who fear either ethical breaches or professional repercussions. The core challenge lies not in detecting AI assistance, but in accurately defining the extent of its contribution to the final manuscript. Authors must distinguish between minor editorial polishing and substantial conceptual generation, a distinction that current guidelines often fail to clarify adequately.

**Also worth reading:** [How Should Self-Published Authors Handle AI Disclosure for Self-Published Books?](https://storywriter.pro/knowledge/how_should_self-published_authors_handle_ai_disclosure_for_self-published_books.php) · [Publisher AI Transparency Checklist: What Every Media Company Must Disclose?](https://storywriter.pro/knowledge/publisher_ai_transparency_checklist_what_every_media_company_must_disclose.php) · [What Must Authors Disclose About Generative AI Use in Journal Submissions?](https://storywriter.pro/knowledge/what_must_authors_disclose_about_generative_ai_use_in_journal_submissions.php)

To resolve this ambiguity, writers should adopt a principle of proactive transparency rather than reactive compliance. Instead of waiting for specific journal mandates, authors ought to disclose any AI interaction in the acknowledgments section, specifying the nature and extent of the assistance provided. This approach aligns with emerging best practices from leading universities and research organizations, which emphasize trust and reproducibility over rigid categorization. By clearly stating how AI tools were utilized, authors empower readers to assess the work’s originality and integrity. Such clarity fosters a culture of accountability, ensuring that technological advancements enhance rather than obscure human scholarly contribution. Ultimately, consistent disclosure builds reader confidence and upholds the foundational values of honest academic communication in an evolving digital era.

## Addressing The AI Slop Problem

Publishers should require disclosure only when AI materially shapes the published work, not when it merely polishes language. A short statement in the methods or acknowledgments should name the tool, describe the task, and confirm human review. Routine spelling, grammar, or formatting assistance can remain undisclosed because it does not change intellectual content. This keeps disclosures meaningful and avoids burying readers in trivial notices while still giving editors enough information to assess responsibility.

For images, charts, or generated figures, disclosure should be mandatory and specific because visual provenance affects trust. Authors should identify whether AI created, altered, or enhanced the image and retain source files for verification. Journals can standardize one sentence rather than inventing separate forms for every submission. Clear thresholds reduce confusion, make compliance practical, and help readers distinguish decoration from evidence without adding unnecessary paperwork to the publication process.

## Handling Oxford's New Disclosure Rules

Oxford’s updated guidance moves disclosure from a vague ethics statement to a practical, task-level note. Authors should name the tool, version, and specific use, such as language editing, code generation, or image creation, while avoiding exhaustive prompt logs. This gives editors enough context to judge contribution and risk without turning the manuscript into an audit trail. It also helps readers understand which intellectual steps remained human.

The unresolved question is where that note belongs. The cleanest answer is to place it in the methods or acknowledgements section, close to the affected content, and mirror it in submission metadata when required. If AI touched only minor wording, a concise sentence suffices; if it shaped analysis, data handling, or visuals, the disclosure should be more detailed and reviewed by all authors before submission. Either way, consistency across the manuscript and portal prevents confusion.

## Managing Cyber Policy Workflow Breaks

Policy revisions that break authorized bug-bounty workflows need an operational fix, not a blame cycle. Publish a dated change log, map each retired rule to a safe-harbor equivalent, and give researchers a named review channel before enforcement. Treat good-faith submissions under prior language as valid if they met the rules at submission time, and record the policy version in every triage report. That keeps security research auditable while legal teams update language.

For disclosure in published work, settle the uncertainty with a materiality and reproducibility test. Authors should disclose AI when it generated or substantially edited text, images, data, code, or analysis, or when it shaped a reported conclusion; they need not disclose incidental spellchecking or formatting. The disclosure should name the tool, version/date, purpose, and human verification, ideally in a methods or acknowledgments note. Journals should apply this baseline consistently rather than requiring blanket declarations, and authors should confirm compliance with the venue’s current policy.

## Clarifying JACC Exemption Standards

The Journal of the American College of Cardiology stands out because it does not require AI disclosure in the way most journals now do. Its editors argue that AI functions as a writing aid comparable to grammar checkers and statistical software, and that the human author retains full responsibility for accuracy and interpretation. This position resolves the disclosure question for a specific category of use: when AI assists with language polishing, formatting, or drafting within content the author has substantively created and verified, no statement is needed.

The practical boundary matters for authors submitting across journals. JACC's exemption does not extend to AI-generated research content, fabricated data, or novel intellectual contributions, which remain prohibited regardless of journal policy. Authors should read each venue's guidelines individually, since Oxford, CASRAI, and others draw the line differently. The safest interpretation is that JACC exempts mechanical assistance, not substantive generation, and treating that distinction as the floor elsewhere prevents accidental misconduct.

## Disclosure Standards Across Publishers

| Policy source | Disclosure standard | Author action |
| --- | --- | --- |
| Oxford University Press | Updated academic-author guidance emphasizes transparent disclosure of generative AI use rather than blanket prohibition. | State the tool, purpose, scope, and human accountability in the location required by the journal. |
| JACC Journals | Does not require disclosure for AI in medical writing in the cited guidance, while accuracy and authorship rules still apply. | Verify article type and current policy; disclose substantive AI-generated content if it affects claims or accountability. |
| CASRAI / journal image rules | AI-generated or edited images require specific provenance and disclosure attention. | Label AI-generated or altered images, describe the process, and confirm permissions and journal requirements. |
| Cross-publisher practical baseline | Confusion persists, so clear, specific disclosure is safer than vague statements. | Use a concise statement covering tool, task, extent, verification, and author responsibility; follow the target journal. |

Across these policies, the safest disclosure identifies the tool, the task, the extent of use, and the human review applied. Oxford-style guidance rewards confidence and specificity; JACC’s narrower stance does not erase journal-specific or image-provenance duties. Authors should place a concise statement where readers expect it—methods, acknowledgments, or figure legends—and confirm the target journal’s current instructions before submission.

## Quick answers

### What constitutes the AI shovelware problem?

It refers to low-effort, AI-generated content flooding platforms and degrading user experience.

### Why do JACC authors skip disclosure?

JACC does not require disclosure because AI tools are not considered authors of medical writing.

### How do cyber policy changes affect bounties?

Opus 4.6/4.7 updates break authorized workflows, requiring researchers to verify new compliance steps.

### What percentage of researchers lack clarity?

Three-quarters of researchers remain unsure about how to properly disclose AI usage in studies.

Canonical: https://storywriter.pro/knowledge/how_to_disclose_ai_use_in_published_work.php
Markdown: https://storywriter.pro/knowledge/how_to_disclose_ai_use_in_published_work.php/index.md