Purpose, Scope and Durability
This policy governs the use of artificial intelligence — including generative and assistive systems — by authors, reviewers, editors, staff and the publisher in work connected with LMI journals. It is stated in terms of function and risk rather than named products, and applies to technologies not yet in existence at the time of adoption. LMI shall review it periodically as capabilities change.
For this framework, artificial intelligence means computational systems performing tasks that would otherwise require human judgement or authorship; generative artificial intelligence means such systems that produce text, images, code, data or other content.
Authors
An AI system cannot be an author. Authorship requires accountability, agreement to be accountable, and the capacity to respond to questions about the work (C1) — none of which a system can bear. AI tools shall not be listed as authors or co-authors.
Human authors remain fully accountable for everything they submit, including any part produced with AI assistance. Accordingly, authors shall:
- Disclose substantive use. Where generative AI materially contributed to the content of the work — drafting substantive text, generating images or figures, producing code, analysing or generating data — the use shall be disclosed at submission and in the manuscript or published item where appropriate, identifying the system or class of system, what it was used for, and the human checks applied. Routine assistance that does not alter substantive content, such as spelling, grammar, reference formatting or translation checking, does not require equivalent disclosure unless the journal’s instructions state otherwise. AI-generated output shall not be cited as the primary evidentiary source for a factual or scholarly claim.
- Not present generated material as genuine observation. AI-generated or AI-altered data, images, results, quotations or references shall not be presented as authentic research output. Image-integrity standards for genuine images are governed by E3.
- Verify. Authors shall independently verify every factual statement, citation and quotation arising from AI assistance. Fabricated or inaccurate references introduced by AI are the authors' responsibility.
- Respect law and confidentiality. Use of AI shall not infringe copyright or third-party rights (F1), disclose confidential information, or expose personal or participant data (A5, D1).
Reviewers
Reviewer obligations are established in B2 and applied here. A reviewer shall not enter a manuscript under review, its data or the review report into any public or uncontrolled AI service; doing so may constitute a breach of confidentiality under A5. Assessment is the reviewer's own human responsibility and shall not be delegated to a system; any permitted assistive use shall comply with the journal's disclosure requirements and shall not diminish the reviewer's accountability for the content of the review.
Editors and Editorial Staff
Editors and staff shall not enter confidential submissions, review reports, complaint or investigation materials into uncontrolled external systems (A5). AI may support administrative and screening tasks — formatting checks, similarity or image screening, metadata handling, reference validation, triage support — where the system is under appropriate contractual and security control and its output is treated as decision support.
An AI system shall not make, or be represented as making, a decision to accept or reject a manuscript, a determination of misconduct, or a record-action decision. Such decisions require identified human editorial authority (A3, B1) and, for misconduct, the process in C5. Where automated screening contributes to an adverse outcome, the person affected shall be told the substance of the concern, not merely that a tool flagged them (C2).
Publisher Governance
LMI shall govern AI systems it adopts institutionally through: contractual and data-protection safeguards ensuring confidential material is not used to train external models or disclosed beyond the permitted purpose; documented human oversight of any AI-supported process affecting authors, reviewers or content; transparency where AI materially affects how submissions are handled; and periodic review of this policy and of deployed systems.
LMI makes no claim that AI-generated content can be reliably detected. Detection tools may inform inquiry but shall not by themselves establish that content was AI-generated or that any person acted improperly; concerns are pursued under C5 on the ordinary standards of evidence and fair process.
Standards and Guidance
Informed by COPE guidance on artificial intelligence and authorship, the ICMJE Recommendations on AI-assisted technology, and WAME recommendations on chatbots and scholarly manuscripts.