GRHAS permits responsible use of AI tools in research and manuscript preparation while preserving human authorship, source accuracy and confidential assessment. Authors should explain substantive AI assistance and provide enough information to evaluate any AI method on which the findings depend.

Human responsibility

Authors are responsible for the reasoning, evidence, factual claims, references and final wording of their article. AI systems cannot accept accountability or approve a submission and cannot be listed as authors. Human contributors should fulfil the journal’s authorship requirements.

Check generated references against the original publications and verify quotations, translations and factual assertions. AI-generated material should be examined for bias, unsupported claims and inappropriate reuse of others’ work.

Disclosure in manuscript preparation

Disclose substantive use of generative AI by identifying the tool and describing the task, such as generating a preliminary summary, assisting translation or drafting material later revised by the authors. Explain the human checking and revision that followed.

Routine spelling and grammar checking can be distinguished from tools that generate intellectual content. When a tool materially affects the analysis, argument, translation or interpretation, describe that use in the manuscript. Supply identifying declarations separately during anonymous review where necessary.

AI as a research method

Report the system and version where available, the task, relevant input materials, processing or prompting procedures, validation approach and analytical choices. Explain how outputs were assessed and how the method relates to the research question.

For images, recordings or other synthetic material, identify what was generated and how it functions in the study. Preserve the distinction between a research illustration, an artistic experiment and evidence presented as a record of an event.

Confidentiality and review

Authors should protect personal information, restricted archives and third-party material when using external tools. Obtain the necessary authority before transmitting protected content.

Reviewers and editors must preserve the confidentiality of submissions and review reports. Confidential material should not be uploaded to public generative AI services. Scholarly judgements and recommendations remain the responsibility of the appointed human reviewer and editor.

Integrity concerns

Undisclosed or misleading use is assessed according to its effect on authorship, evidence and the reliability of the work. The editor may request clarification, original materials or a revised declaration. Material problems after publication are handled through the journal’s correction and research-integrity procedures.