Table 1
Participant Demographics.
| DEMOGRAPHIC | # (%) PARTICIPANTS N = 18 |
|---|---|
| Journal type | |
| Health professions education journal | 14 (77) |
| General medical journal | 4 (22) |
| Role | |
| Editor in Chief | 9 (50) |
| Deputy/Associate editor | 9 (50) |
| Participant Country | |
| Australia | 1 (5) |
| Canada | 4 (22) |
| Netherlands | 1 (5) |
| Oman | 1 (5) |
| Singapore | 1 (5) |
| UK | 3 (5) |
| USA | 7 (38) |
| Gender | |
| Female | 10 (55) |
| Male | 8 (44) |
| Years of editor experience | |
| <5 | 1 (5) |
| >10 | 17 (94) |
Table 2
AI-Use Disclosure Guidance for HPE Editors & Journals.
| Understand how your journal’s online architecture shapes AI-use disclosure |
| Review online submission architecture to understand how disclosure is prompted in your system: Is refinement required given speed of change? Understand how disclosures populate from the online system into the editorial process: Who sees disclosures and where in the process? |
| Provide clear, detailed guidance about what your journal expects in AI-use disclosure |
| Review guidance for authors: Is it sufficient? Does it reference principles such as reproducibility in problematic ways? Provide explicit examples to show authors where your journal sets the necessity & sufficiency thresholds: Don’t over simplify – how should authors handle blurry situations? Remind authors to attest to accountability regardless of where the disclosure appears in the manuscript. |
| Educate authors, editorial team and peer reviewers to cultivate safety |
| Educate the editorial team to ensure the co-construction philosophy is universally understood. Make explicit to authors that disclosure is non-punitive. Provide guidance to peer reviewers regarding AI-use disclosure: What is their role? |
Table 3
AI-Use Disclosure Guidance for HPE Authors based on Editor Expectations.
| DISCLOSURE BASICS | COMPLEXITIES OF SUFFICIENCY AND NECESSITY |
|---|---|
| State how you used AI. Example: “We used ChatGPT 4o (May 2024 version) to shorten our abstract and strengthen our study limitations.” | Distinguish between technical & intellectual use. Intellectual uses are of more concern to editors and require more explanation. Example: “We used ChatGPT 4o (May 2024 version) to help brainstorm counterarguments in the Discussion. We prompted counterarguments by discipline (e.g., psychology) and research paradigm (e.g., positivism).” |
| Be specific: tool, tasks, location. Include at minimum: the tool name, version or date (if known), the tasks performed, and the manuscript section(s) affected. | Be strategic about the level of detail. More detail is not always better or feasible. Scale down details for technical use (e.g., editing grammar); scale up for more intellectual use (e.g., creating interview prompts). Editors may request more details as the manuscript nears acceptance. |
| Add a responsibility statement. Attest that you are accountable for the content. Example: “All AI-assisted content was reviewed and verified by the authors, who take full responsibility for its accuracy and integrity.” | Indicate how you verified content. Verification is a process, not a promise. Be transparent about how you verified. Example: “We verified accuracy of AI-produced text by conducting our own PubMed search and reviewing key sources. Where we found discrepancies or oversimplifications, we edited the AI text as needed.” Consider what aspects of AI-use might not be transparent or verifiable. |
| Put the disclosure in your manuscript. Editors prefer this in the Methods or other section as relevant, rather than in Acknowledgements. | Aim for credibility, not necessarily reproducibility. Share your thinking process; acknowledge inconsistencies in AI output; describe how you iterated prompts to achieve better results. If you complete a checkbox or text box during online submission, ensure coherence with manuscript disclosure text. |
| When in doubt, disclose. If the AI use shaped ideas, text structure, or scholarly content—not just grammar—editors generally expect disclosure. | Don’t avoid disclosure in fear of punitive action. Journals and authors are co-constructing the norms of disclosure, learning together in a fast-changing situation as AI technology and literacy develop. Disclose in this spirit. Provide detail; editors can request unnecessary details be removed upon acceptance. |