Disclosure
Material generative-AI use in manuscript preparation should be disclosed transparently. Routine spelling or grammar assistance need not receive the same detailed disclosure unless a journal states otherwise.
Verification of Claims and References
Authors must independently verify AI-assisted text, calculations, references, quotations, summaries, and interpretations against reliable sources and the underlying evidence. Plausible-looking output is not verification; fabricated, mismatched, or non-existent citations must not be submitted.
Prohibited Fabrication and Misrepresentation
AI must not fabricate patients, participants, clinical or laboratory data, references, results, images, diagnostic findings, experiments, adverse events, or outcomes. AI-generated or modified images or data must not be presented as authentic research evidence when they are not.
AI-generated and AI-modified Images
AI-generated or materially AI-modified imagery must not be represented as authentic microscopy, radiology, pathology, clinical photography, participants, specimens, or experimental evidence. Legitimate analytical or illustrative use must be transparent and must not obscure provenance or mislead readers.
Editorial and Review Confidentiality
Editors and reviewers must protect confidentiality. Confidential manuscripts, patient information, unpublished results, reviewer reports, or protected information must not be uploaded to external generative-AI services where that could compromise confidentiality, rights, or data protection. Convenience does not override duties attached to privileged material.