Use of AI & Automated Tools

3.1 Scope
This policy covers any artificial intelligence (AI) or automated system used in the creation, analysis, visualization, translation, or editing of scholarly work, as well as tools used by editors and reviewers during assessment. It applies to text, images, audio/video, code, and data.

3.2 AI Is Not an Author

AI systems-generative or otherwise-cannot be listed as authors. Authorship requires human accountability for the integrity, originality, and accuracy of the work. Human authors must accept responsibility for all content generated or assisted by AI.

3.3 Disclosure Requirement (Manuscripts)

Authors must disclose, in the manuscript, any use of AI or automated tools with specific detail:

  • What tool was used (product name and version, if known).
  • Where it was used (e.g., drafting text, grammar polishing, figure generation, coding assistance, statistical analysis, translation).
  • How outputs were verified and edited by the authors.
  • Data exposure: whether any non-public or personal data were input into the tool.
  • A short AI Use Statement appears in the Methods or Acknowledgments. If no AI was used, no statement is required.

3.4 Acceptable Uses (with human oversight)

  • Language editing (grammar, clarity, translation).
  • Drafting boilerplate texts (e.g., method templates) that are fully reviewed and rewritten as needed by authors.
  • Code assistance for routine tasks, provided authors review, test, and document code and dependencies.
  • Figure creation or enhancement that does not misrepresent data; all manipulations must be disclosed.
  • Screening utilities (e.g., reference formatting, duplicate-checking) that do not alter scientific content.

3.5 Prohibited Uses

  • Listing AI as an author or citing an AI system as a primary source of facts or original ideas.
  • Using AI to fabricate, falsify, or inappropriately manipulate data, images, statistics, or citations (including hallucinated references).
  • Concealing AI-generated content or failing to disclose meaningful AI assistance.
  • Feeding confidential or identifiable participant data to external tools without lawful basis, consent (where required), and adequate safeguards.
  • Using AI to write peer review reports on manuscripts not assigned to the user or to breach review confidentiality.

3.6 Verification & Accountability (Authors)

  • Authors must fact-check AI-assisted text and validate any numbers, code outputs, or statistical results.
  • Authors remain responsible for plagiarism checks, permissions for third-party content, and accurate citation of all sources.
  • For code or analysis produced with AI, authors must provide runnable code, details of the environment (versions, packages), and test evidence that results are reproducible.
  • Image outputs must be traceable to original data; keep unprocessed originals and processing logs.

3.7 Data Protection & Confidentiality

  • Do not upload personal data, confidential data, or unpublished manuscripts to tools that may reuse inputs for training or lack adequate contractual safeguards.
  • When institutional or legal restrictions apply (e.g., data transfer limits), authors must comply and document their compliance in the Data Availability or Methods section.
  • Redact or anonymize data before tool use when feasible.

3.8 Citation & Permissions

  • Cite datasets, code libraries, and software used in analysis per field norms.
  • Do not cite a chat transcript as a scholarly source; instead, cite the underlying primary literature the model references.
  • If AI-generated images or media are used, disclose their status and ensure no third-party rights are infringed.

3.9 Image, Figure, and Media Integrity

  • Disclose all adjustments (e.g., brightness, contrast, denoising) and ensure they are uniform and non-misleading.
  • AI-based upscaling, inpainting, or segmentation must not alter the scientific meaning of images. Provide original files upon request.
  • For synthetic illustrations (e.g., conceptual diagrams), label as illustration and avoid implying empirical measurement.

3.10 Statistical, Computational, and Code Assistance

  • Document prompts, parameters, and post-processing steps that materially affect results.
  • Provide a Data/Code Availability statement with persistent links (e.g., DOI repositories).
  • For AI-assisted code, include unit tests or minimal reproducible examples that demonstrate correctness.

3.11 Editors & Reviewers — Responsible Use

Editors and reviewers may use AI for language polishing of their own text or for organizational aids (e.g., summarizing public methods), not for confidential manuscript content unless:

  • The tool is approved for confidential use, and no manuscript text or data leave secure environments.
  • Reviews must reflect the reviewer’s own critical judgment; do not submit AI-generated reviews as if they were original assessments.
  • Do not upload confidential manuscripts or reviewer reports to tools that store or train on user inputs.

3.12 Publisher Screening Tools

The publisher may employ vetted automated tools to assist with:

  • Plagiarism and duplicate publication checks,
  • Image integrity screening,
  • Reference and citation anomaly detection,
  • Statistical or code red flags (where appropriate).
  • Findings trigger human editorial review; automated outputs do not determine decisions on their own.

3.13 Disclosure Format (Template)

AI Use Statement (example):
“Portions of language editing were assisted by an AI tool. The authors verified and edited all content. No confidential or personal data were shared with external systems. Statistical analyses and figures were performed by the authors; code and data are available as stated in the Data Availability section.”

3.14 Record-Keeping & Audit

  • Keep original data, code, images, and AI-related logs (e.g., prompts, major parameters) for a reasonable retention period consistent with field norms and legal requirements.
  • Upon editorial request, provide materials sufficient to verify claims. Failure to verify may lead to correction or retraction.

3.15 Breaches & Remedies

  • Non-disclosure, misrepresentation, or misuse of AI may result in rejection, correction, expression of concern, or retraction.
  • Serious or repeated breaches may be referred to institutions or funders and may incur submission bans for a defined period.

3.16 Equity & Accessibility

Authors who use AI for accessibility (e.g., dyslexia-friendly drafting, language support) are welcome to do so with disclosure and verification. Accessibility support never replaces author responsibility.

3.17 Training Data Ethics (Awareness)

When AI tools are central to the research (e.g., new models), authors must describe training data provenance, licensing, consent where applicable, risk of bias, and evaluation methods. Provide a model card or equivalent documentation.