Contents: 1 Purpose and scope · 2 Definitions · 3 General principles · 4 Use of generative AI by authors · 5 Generative AI and peer review · 6 Generative AI and editors
1 Purpose and scope
This policy governs generative AI and AI-assisted technologies used in manuscript preparation, legal research, submission screening, peer review, editorial administration and publication. It applies to authors, reviewers, editors, guest editors, production staff and others acting for LEXANTARA. Its purpose is to preserve accountable human scholarship, verifiable legal analysis, confidentiality and transparency while allowing responsible assistance. No AI vendor is approved by this policy.
2 Definitions
Generative AI. Systems that produce or substantially transform text, code, images, audio, video or other content in response to prompts or other inputs.
AI-assisted technologies. Tools using AI to support search, translation, summarisation, transcription, classification, analysis or editorial work. Their actual function, not merely their product name, determines reporting requirements.
Substantive use. Assistance that materially affects scholarly wording, argument, translation, source selection, analysis, evidence or interpretation. Basic spelling, punctuation and grammar correction that does not change substantive meaning is routine assistance.
AI as method or object of research. The use or examination of an AI system as part of a study, distinct from using a tool to draft its report. It requires appropriate methodological reporting even if manuscript drafting is entirely human.
Confidential material. Unpublished manuscripts, reports, correspondence, identifying metadata, restricted data, privileged advice, client records, confidential awards and settlement material not authorised for external disclosure.
3 General principles
Human authors and decision-makers remain accountable. AI cannot be an author, reviewer or acceptance authority. AI output must be checked against real sources; it cannot establish that a source exists, a legal rule is in force or a conclusion is valid.
Use must respect intellectual property, informed consent, privacy, confidentiality and professional obligations. Assess bias and limitations, disclose substantive assistance, and never fabricate evidence or conceal plagiarism. A private, paid or institutionally supplied account does not itself establish adequate safeguards.
Every manuscript includes a declaration of substantive AI use or of no substantive AI use. Routine basic corrections need not be itemised. Research-related AI use is also explained in the method section. No misconduct decision rests solely on an AI-detector score; the editor applies evidence-based assessment and due process.
4 Use of generative AI by authors
Permitted assistance and accountability
LEXANTARA permits responsible AI-assisted research and writing when authors retain intellectual control, protect rights and confidentiality, report methods adequately and disclose substantive use. AI assistance does not replace an original contribution, sound evidence or author accountability. Tools cannot satisfy authorship responsibilities and must not appear in the author list.
Permitted uses may include language revision, translation, coding assistance, exploratory analysis, literature discovery and preparation of illustrative material, subject to the safeguards below. The editor may seek sufficient documentation to assess validity. The policy applies to generative text, code, image, audio and video tools and to AI systems used in research workflows.
A legal source does not exist merely because an AI tool supplies a plausible citation. For every cited legal source, verify the jurisdiction, court or authority, date, operative status, quotation, translation, pinpoint reference and authenticity against the actual source. Check whether translations are official and state the relevant legal cutoff date. Never upload privileged advice, client files, confidential awards or settlement material without a lawful basis, the necessary permission and verified safeguards.
Disclosure requirements
Authors must disclose substantive AI use in a dedicated declaration before the references. Identify the provider, tool and model or version where available; when it was used; the purpose; the affected manuscript components; and the human checks performed. Substantive AI-assisted translation and drafting require disclosure. Routine spelling, punctuation, reference formatting or basic grammar correction need not be declared unless substantive meaning is changed.
Every submission must include either a substantive-use declaration or a no-substantive-use declaration. If substantive AI is used after revision or acceptance, update the declaration and notify the handling editor.
When AI affects evidence generation, coding, transcription, classification, analysis or reported results, describe it in the Methods section as well as in the declaration. Report settings, validation, uncertainty and relevant prompts or workflow details sufficient for evaluation, subject to consent, safety, licensing and privacy constraints. Retain records and explain any justified limits to sharing. A generic acknowledgement does not substitute for methodological reporting.
Verification and prohibited practices
Verify citations against the actual source and check quotations, numerical claims, translations, code and statistical outputs. AI output is not evidence that a cited source exists or supports a claim.
Do not fabricate participants, interviews, field observations, datasets, quotations, findings, sources or approvals. Synthetic or simulated material is acceptable only as a transparently labelled research method with a justified design; it must never be presented as observed evidence. Hypothetical legal scenarios must be labelled and must not be represented as actual disputes, judgments or transactions.
Do not use AI to conceal plagiarism, impersonate authors or reviewers, manipulate peer review, or create misleading figures or audiovisual evidence. AI-generated figures must be disclosed, accurately described and rights-cleared, and distinguished from empirical material. Images presented as photographs or authentic documentary evidence must not be generated or altered to mislead.
Do not upload confidential manuscripts, restricted data or identifiable participant information without an appropriate lawful and ethical basis, necessary permissions and verified safeguards. A paid or private account alone is not proof of adequate protection. AI-detection scores alone cannot establish misconduct or justify rejection or retraction.
Example declarations
Substantive use: During preparation of this article, the authors used [provider, tool, model/version, date] for [specific task and sections]. The authors checked [sources, translations, code or outputs] by [verification procedure], revised the material and accept responsibility for the final work. Research-related use is described in Methods.
No substantive use: The authors declare that no generative AI was used for substantive drafting, research design, evidence generation or analysis. Routine spelling or formatting assistance, if any, did not alter substantive content.
5 Generative AI and peer review
Reviewers must personally evaluate the manuscript and write their scholarly assessment. They must not send confidential submissions, extracts, figures, supplements, reports, correspondence or identifying information to external generative AI tools, nor use them to draft, rewrite or recommend a scholarly review from confidential material. Removing an author’s name does not remove confidentiality.
No AI-assisted scholarly review is currently authorised. Any future limited assistance requires prior documented journal authorisation for the specific function, verified protection against unauthorised retention or training, appropriate permissions and human accountability. Any authorised use must be disclosed confidentially to the editor, naming the tool, task and material processed, and authors receive transparent information about the use of their material.
Reviewers may consult public guidance on review practice without disclosing the submission. If an AI-generated or fabricated report is suspected, editors verify the concern and seek clarification rather than relying on detector scores. Unauthorised exposure must be reported promptly to the Managing Editor and the privacy contact, with processing stopped and relevant records preserved.
6 Generative AI and editors
Default rule
Editorial screening, reviewer selection, peer review and final decisions are human-led. Editors and reviewers must not upload confidential manuscripts, reports, correspondence, unpublished data or identifying metadata to external generative AI services, or use generative AI to draft scholarly assessments or recommend decisions from confidential material.
Non-generative administrative tools, such as reference checks, formatting utilities and similarity screening, may assist staff only after appropriate confidentiality and security checks. Similarity outputs require human interpretation. Any system that processes content must be assessed for retention, access, data sharing, security and model-training terms. No AI system or screening vendor is currently approved.
Controlled changes and disclosure
Any future exception requires documented publisher authorisation and editorial approval before use, a risk assessment, verification that confidential content is protected and not reused for unauthorised training, appropriate permissions, an accountable human operator, and an updated public policy naming the function and tool. Affected authors and reviewers receive clear information, with consent where required. Secure or institutional hosting does not remove disclosure or human-accountability requirements.
Research assessments and decisions remain the responsibility of named human editors and reviewers; tools cannot serve as reviewers. Reviewer invitations and selection remain human-approved even where an authorised discovery tool suggests names.
Automated detection and due process
AI-detection results are not treated as proof of AI use, authorship fraud or misconduct, and no automatic rejection or retraction is based solely on a score. Editors assess specific concerns, request relevant explanations or documentation and consider language and disciplinary context. Disclosure omissions are assessed proportionately; responsible disclosed assistance is not itself misconduct.
Security incidents and training
A person discovering an unauthorised disclosure must notify the Managing Editor and the designated privacy contact promptly, halt further exposure and preserve relevant records. The publisher assesses containment, provider deletion requests where feasible, participant and author risks, and notification duties under applicable requirements. Independent handling is required if an editor is implicated. The journal does not promise deletion or confidentiality outcomes that cannot be verified.
Editors, guest editors and reviewers receive induction on this policy before access to live submissions. The publisher reviews approved systems and controls at least annually, keeps an internal register of authorised tools, purposes and safeguards, and this page describes actual practice.
Content mining and model training
Reuse of published articles follows the CC BY 4.0 licence and applicable law; the journal does not promise blanket restrictions inconsistent with that licence. Any separate content-mining or AI-company arrangement will be disclosed. Unpublished submissions, reviewer reports, confidential editorial records and participant data are not made available for training through acceptance of a submission.
Policy version: founding draft 2.2 (2 November 2026), pending formal adoption.