Generative AI Policy

Generative artificial-intelligence tools are increasingly used in scholarly and clinical writing. BJHMR permits their responsible use while protecting the accountability, originality and confidentiality on which trust in medical research depends. This policy is consistent with the positions of COPE and the ICMJE and with emerging practice across reputable medical publishers.

Authorship and accountability

A generative-AI tool cannot be an author, since it cannot take responsibility for a work or be accountable for it. Such tools must never be listed as authors, and the human authors remain wholly responsible for all content, including any text, analysis or figures produced with AI assistance, and for its accuracy.

Disclosure and prohibited uses

Authors must disclose any substantive use of generative AI, naming the tool and version and describing how it was used; routine tools such as spelling and grammar checkers do not require disclosure. Authors must verify all AI-assisted output, since these tools can generate plausible but erroneous or fabricated results, references or clinical statements. Using AI to fabricate or falsify data, results or references, or to manipulate images, constitutes research misconduct. AI must not be used to generate or alter clinical images or data in ways that misrepresent the research.

Reviewers and editors

Reviewers must not upload any part of a manuscript to a generative-AI tool, as this breaches the confidentiality of unpublished clinical research, and must not delegate their assessment to such tools; editors must likewise protect the confidentiality and integrity of the process. The journal will keep this policy under review as practice develops.