This policy applies when artificial intelligence is an object of research, forms part of a research method, processes or analyzes data, generates or modifies research materials, assists manuscript preparation or revision, supports peer review or editorial assessment, or is used during production and publication.

• Human accountability: qualified humans remain responsible for every judgment and submitted output.
• Transparency: substantive AI use must be disclosed accurately.
• Accuracy and originality: AI-assisted content must be verified and must not conceal plagiarism or misrepresent its origin.
• Confidentiality and privacy: protected information must not be submitted to an unauthorized AI system.
• Fairness: users must assess bias, discrimination, exclusion, and unequal performance.
• Reproducibility: material information about systems, prompts, settings, and procedures should be reported sufficiently for evaluation.
• Intellectual property: copyright, licensing, database, authorship, and third-party rights must be respected.
• Research integrity: AI must not fabricate, falsify, conceal, or deceptively manipulate scholarly evidence.

Use of an AI tool does not transfer responsibility to the developer, provider, or system. Final scientific, methodological, ethical, editorial, and publication decisions must be made by qualified humans.

An AI system, chatbot, large language model, or automated tool must not be listed as an author, co-author, corresponding author, group author, or contributor capable of accepting final responsibility.

AI systems cannot approve the final manuscript, consent to submission, accept accountability, guarantee accuracy or originality, disclose conflicts, respond independently to editors, cooperate with investigations, hold copyright, or enter licensing arrangements.

Human authors remain responsible for all content generated, revised, translated, classified, analyzed, visualized, or otherwise assisted by AI.

Authors using AI must have a legitimate scholarly purpose, understand relevant limitations, verify every material output, correct inaccuracies and bias, protect confidential information, respect intellectual-property obligations, comply with ethics approval and consent, identify synthetic materials, preserve appropriate documentation, and accept full responsibility for the final work.

• verify every citation, DOI, quotation, numerical claim, calculation, translation, image, code output, and analytical result;
• explain why AI was used and which tasks it performed;
• describe how outputs, errors, bias, and limitations were assessed;
• maintain meaningful human oversight and make final decisions independently; and
• explain how AI use affected interpretation, validity, or reproducibility.

Substantive AI use must be disclosed even if authors later revise the output extensively.

• generating or substantially rewriting text;
• translating scholarly or participant content;
• literature searching, screening, classification, extraction, or evidence synthesis;
• hypothesis, research-question, method, instrument, or material development;
• code generation, completion, debugging, or documentation;
• data cleaning, transformation, analysis, statistics, qualitative coding, or thematic analysis;
• transcription, labeling, classification, or social media analysis;
• generation or substantial modification of images, figures, tables, or synthetic data;
• machine-learning model development or evaluation;
• automated participant interactions or material prompt-based procedures; and
• substantial drafting of responses to reviewers or another task affecting meaning, accuracy, originality, analysis, interpretation, or presentation.

Detailed disclosure is not normally required for routine spelling, punctuation, basic grammar checking, reference formatting, file conversion, ordinary word processing, noninterpretive calculation, or formatting that does not alter scholarly content.

This exception applies only when the tool does not generate intellectual content, substantially rewrite text, alter meaning, select or interpret evidence, generate references, make analytical decisions, modify research evidence, or introduce material content. When uncertain, authors should disclose the use.

A proportionate disclosure should identify the tool and provider, model and version where available, access date or period, purpose, affected manuscript sections, data or materials processed, material prompts or settings, verification procedures, human revision, and known reproducibility limitations.

Substantive AI use should be disclosed in the cover letter and Title Page, in the Methods section when AI formed part of the research method, in a dedicated Artificial Intelligence Disclosure section when AI assisted preparation, and in captions or supplementary materials where relevant. Information needed by readers must not appear only in confidential correspondence.

AI systems may produce authoritative-sounding but incorrect, incomplete, biased, or fabricated output. Authors must independently verify facts, numbers, quotations, calculations, translations, code, statistics, images, provenance, and consistency with underlying evidence.

An AI-generated reference must not be included unless the authors have confirmed that the source exists, bibliographic information and DOI or URL are accurate, the original source was consulted, and the source supports the associated claim.

AI-assisted wording does not remove citation obligations. AI must not be used to paraphrase copied material to avoid detection, lower similarity scores artificially, conceal the origin of ideas, imitate another author improperly, or present translated plagiarism as original work.

• fabricating participants, data, observations, experiments, responses, interviews, quotations, social media content, clinical or institutional records, or statistical results;
• creating false references, DOI numbers, ethics approvals, consent forms, registrations, reviewer identities, or review reports;
• concealing missing or contradictory data or manipulating findings;
• creating deceptive empirical images, screenshots, or research records;
• impersonating another person or concealing authorship, paper-mill activity, or third-party involvement;
• producing false responses to reviewers or falsely claiming that an analysis, experiment, revision, or correction was completed; and
• misrepresenting synthetic or AI-generated content as real-world empirical evidence.

Such conduct may result in rejection, correction, Expression of Concern, retraction, institutional referral, or another action under applicable integrity and publication policies.

AI-assisted systematic reviews, scoping reviews, meta-analyses, and other evidence syntheses should report the tool, model, version, task, affected stage, prompts or rules, validation, human review, disagreement handling, and final human responsibility. AI must not make unverified final decisions on eligibility, risk of bias, extraction, certainty, statistical interpretation, or conclusions.

AI-assisted transcription, translation, coding, categorization, or thematic analysis must be disclosed. Authors should describe the supplied data, privacy protections, system, task, verification, cultural and linguistic context, disagreement handling, bias and loss-of-meaning risks, influence on themes, and whether consent and ethics approval permitted the use.

AI-generated themes or classifications must not replace reflexive, contextual, and methodologically appropriate human interpretation. Identifiable participant content must not be uploaded to an unauthorized system.

AI-generated or substantially AI-modified images, figures, graphs, and tables must be identified clearly. AI must not generate or alter content presented as authentic participant photographs, clinical images, laboratory or field observations, experimental outputs, social media screenshots, historical records, documents, or other empirical evidence.

AI-assisted modification of genuine research images must not add or remove features, conceal anomalies, alter measurements, create false observations, change meaning, or misrepresent original evidence. Graphs and tables must reflect genuine verified data, accurate labels and scales, appropriate uncertainty, and human verification.

Authors must retain original files and document material modifications. Conceptual or explanatory AI-generated visuals may be considered when clearly identified, not presented as empirical evidence, not misleading about real individuals, and compliant with ethical and intellectual-property requirements.

Authors using AI for programming, debugging, optimization, software development, or statistics must review and test outputs, verify mathematical and statistical logic, assess security and bias, examine dependencies and licenses, retain final code, validate outputs, report material settings and limitations, and make all final analytical decisions independently.

Reporting should identify the task, input data, preprocessing, model or system, relevant parameters, validation, human oversight, sensitivity or robustness checks, and known limitations. AI-generated code must not be assumed to be correct, secure, unbiased, original, or fit for purpose.

AI-based social media research should address provenance, platform conditions, public or private context, privacy expectations, identifiability, cultural and language variation, bot detection, demographic and model bias, misclassification, sentiment limitations, sensitive inference, profiling, automated labeling, human validation, reidentification, and potential harm.

AI-generated classifications must not be treated automatically as verified facts about identity, health, beliefs, location, vulnerability, or behavior.

Synthetic data must be labeled clearly and accompanied by the purpose, any real data used, model and version, generation procedure, parameters, validation, similarity to real records, privacy safeguards, disclosure risk, bias, limitations, and available code. Synthetic data are not automatically anonymous and must never be presented as observations of real people, platforms, institutions, experiments, or environments.

When an AI system is studied directly, authors should report the provider, model, version, access method and date, account type where relevant, available system instructions, prompts, interaction sequence, parameters, number of generations, selection criteria, safety filters, human editing, exclusions, evaluation procedures, comparison systems, and reproducibility limitations.

Prompts and material interaction records should be retained and shared when ethically, legally, contractually, and technically appropriate. Records may include prompt order, responses, regeneration attempts, settings, excluded outputs, selection decisions, dates, and human revisions.

Prompts or outputs containing personal data, confidential participant information, copyrighted full text, private social media content, proprietary information, security-sensitive information, unlawful content, or other restricted material must not be shared publicly. When full prompts cannot be provided, authors should explain the restriction and provide sufficient methodological detail.

Personal, confidential, unpublished, restricted, embargoed, or identifiable information must not be submitted to an AI system unless law, ethics approval, consent, institutional authorization, contracts, platform terms, security safeguards, retention conditions, and model-training conditions permit the use and relevant risks have been assessed.

Removing names may not make information safe when reidentification remains possible. Protected material includes health, genetic and biometric data, private messages, closed-group content, transcripts, recordings, institutional and commercial records, unpublished research, full-text publications, peer-review material, and information governed by data-use agreements.

Users should review system terms, privacy and retention policies, model-training conditions, confidentiality provisions, copyright, ownership, and licensing. Authors must possess the rights needed to publish AI-assisted content and must ensure the system does not transfer unauthorized rights, expose protected information, prevent lawful publication, or conflict with journal, funder, institutional, or third-party requirements.

Authors must consider whether an AI system reproduces or amplifies bias involving language, geography, disability, age, socioeconomic conditions, institutional prestige, publication language, discipline, data availability, underrepresented populations, or another relevant characteristic.

When AI materially affects classification, prediction, recommendation, or generation, authors should report represented and underrepresented populations, fairness assessment, differential performance, error rates, limitations, mitigation, and effects on interpretation. Computational output must not be described as neutral merely because it was generated automatically.

JSOMER does not treat an AI-detection score as conclusive proof of AI use or misconduct. Detection tools may produce false positives and negatives and may perform differently across languages. A score may prompt further assessment but cannot be the sole basis for a finding.

Reviewers remain personally responsible for the quality, accuracy, fairness, confidentiality, and scholarly substance of their reports.

• Do not upload confidential manuscripts, supplements, or instructions to unauthorized AI systems.
• Do not delegate peer review to AI or submit an AI-generated report without personal scholarly evaluation.
• Do not use AI to identify anonymized authors, fabricate references, or exploit confidential information.
• If an authorized AI tool is used substantively, disclose the use to the handling editor, verify all output, protect confidentiality, prevent model-training use, and make the final recommendation independently.

Editors must not delegate acceptance, rejection, revision, correction, retraction, or another substantive decision to an AI system. Confidential manuscripts, reports, identities, and correspondence must not be uploaded to unauthorized systems, and AI must not be used to identify anonymized participants in review or to replace scientific judgment.

AI may assist with limited administrative classification, reference verification, reviewer discovery, integrity alerts, metadata, format checks, accessibility, or workflow management only with authorization, confidentiality, verification, bias assessment, access control, transparency where relevant, and final human decision-making.

Editorial Board members and Guest Editors follow the same requirements. They must not use confidential submissions to train or test AI, allow AI to favor specified contributors or institutions, or rely on AI-generated assessments without human evaluation.

The publisher, staff, and consultants may use authorized AI-assisted systems for legitimate administrative or production purposes when confidentiality, lawful personal-data processing, security, verification, human oversight, bias assessment, and editorial independence are protected.

Technical automation must not replace scholarly judgment or influence outcomes for financial, personal, institutional, reputational, citation, or indexing reasons. AI-assisted production must not alter scientific meaning, results, quotations, conclusions, ethics statements, conflicts, data availability, copyright, or licensing without author and editorial review.

Undisclosed substantive AI use by authors may lead to clarification, revised disclosure, requests for prompts, outputs, drafts, histories, references, data, images, or code, additional review, suspension, rejection, reversal of acceptance, institutional referral, or post-publication action.

Unauthorized AI use by reviewers, editors, or staff may lead to disregard of a review, reassignment, repeated assessment, restricted future roles, removal from duties, institutional notification, author notification, data-protection response, security review, or another protective measure.

The response depends on materiality, effect on the work, accuracy, confidentiality, privacy, originality, integrity, explanation, and whether transparent correction is possible. An AI-detection result alone is insufficient to establish misconduct.

When a credible AI-related concern arises, JSOMER may suspend evaluation or publication, preserve records, request explanations and supporting materials, verify references and evidence, seek independent technical advice, require revised disclosure or correction, repeat review, reject or withdraw a manuscript, contact an institution, or take another action under the Research Integrity Policy.

Authors will normally have an opportunity to respond. A correction may be appropriate when omitted disclosure can be remedied and the work remains reliable. Retraction may be appropriate when AI was used to fabricate data or people, create false research records or deceptive empirical images, produce false references central to the article, conceal substantial plagiarism, manipulate review, manufacture a paper, or otherwise make findings unreliable.

AI processing of human participant data must be covered by appropriate ethics approval, consent, institutional authorization, data-transfer safeguards, and protection against retention, model training, profiling, automated decisions, reidentification, or harm. Approval for original data collection does not automatically authorize every later AI use.

Relevant data, code, prompts, parameters, materials, and outputs should be shared when ethically, legally, contractually, and technically possible. Proprietary or changing systems may limit reproducibility; such limits must be reported transparently.

AI-assisted content must be publishable lawfully, comply with tool terms, respect third-party rights, and permit application of appropriate journal licenses. AI does not excuse plagiarism, fabrication, falsification, or other integrity failures.

AI-related decisions may be appealed under the Complaints and Appeals Policy when evidence was overlooked, policy was applied incorrectly, a factual error occurred, a detection score was misinterpreted, procedure was materially unfair, or new evidence is available.

Because AI develops rapidly, JSOMER reviews this policy periodically for consistency with actual practice, technological developments, publication ethics, privacy, data protection, copyright, licensing, research integrity, peer-review confidentiality, related journal policies, and relevant scholarly-service requirements. Material revisions will be disclosed on the journal website.