Skip to main content

Updated Guidance for Artificial Intelligence Use in Medical Publications

jjcush@gmail.com
Aug 12, 2026 10:00 am

Know-it-now

  • AI must not be used to generate References.
  • AI should not be used to generate Opinion pieces, Letters to the Editor, and online Comments. These formats now require human-originated insight.
  • Clinical images, illustrations, video, and audio created/manipulated by AI are prohibited outside formally disclosed research methodology.
  • Disclosure of AI use (where appropriate) is specific and mandatory (and placed in the Acknowledgment section).
  • Peer review confidentiality is violated with AI use. Manuscripts or reviewer comments must never be entered into external AI/LLM platforms.

JAMA has updated its guidance for the acceptable use of artificial intelligence (AI) in authoring a medical publication. Despite growing AI use, very low rates of disclosure of AI use have been reported. This update provides specific information about what is permitted and not permitted, and what needs to be disclosed.

Journals have adopted policies regarding appropriate and responsible AI, putting guidelines on use and how and when to disclose AI use. But some authors have avoided disclosure due to fear of biased editorial reviews or unfamiliarity with such disclosures. For those engaged in manuscript writing, peer review, editorial work, or CME/educational content, this policy update has direct, practical implications.

Editors have noted a rising number of AI-generated submissions with inaccurate, fabricated or entirely nonexistent citations that look plausible and at times use real author names and titles that do not correspond to actual papers. Editors have noted a surge in either Opinion pieces and Letters to the Editor, often from authors without topic expertise, and presumably assisted by AI drafting.

Key Updates – these differ according to how or when AI may be employed.

  • References: AI/LLMs must not be used to generate or format bibliographic references; standard reference managers are recommended instead, and authors remain fully responsible for reference accuracy and integrity.
  • Opinion content: Use of AI to draft Opinion manuscripts, Letters to the Editor, and online Comments is no longer permitted; human insight is prioritized for these formats.
  • Clinical images/media: AI-created or -manipulated clinical images, illustrations, video, and audio are prohibited unless part of formally disclosed research methodology. Uniform brightness/contrast adjustments and solid-color background replacement remain acceptable if disclosed with software name/version, and original unedited images must be retained for review.
  • Research and data visualization: AI use is permitted in research methods and in generating graphic visualizations of results, provided it is fully described (Methods section for research; figure legend for visualizations) with software name and version disclosed.
  • Manuscript preparation, literature search, translation: Permitted with disclosure in the Acknowledgment section, provided authors verify accuracy and accept full responsibility for content.
  • Grammar/spelling checks: AI assistance is permitted with no disclosure required.
  • Peer review: Entering a manuscript or reviewer summary into an external AI tool remains prohibited as a confidentiality violation. AI-assisted author responses to reviewers/editors are also not permitted.

Disclosure Requirements. When AI use is reportable, the Acknowledgment section must include software name, version/extension number, manufacturer, dates of use, a description of how and where AI was applied, and confirmation that authors take full responsibility for accuracy and integrity of the generated content.

For authors and reviewers, the practical takeaway is narrow but firm - AI is a permitted research and drafting aid when fully disclosed and verified, but it is now explicitly barred from touching your references, your Opinion/Letters content, your clinical images, and any peer review correspondence. GIt is best to to double-check reference lists in any AI-assisted content or submitted manuscripts.

ADD THE FIRST COMMENT

If you are a health practitioner, you may to comment.

Due to the nature of these comment forums, only health practitioners are allowed to comment at this time.

Disclosures

Disclosures
The author has no conflicts of interest to disclose related to this subject
The author used AI to research and organize this content, and maintains responsibility for its accuracy
×