CONTENTS
- Scope and applicability
- Governing principles
- AI tools and authorship
- Permitted and prohibited uses of AI by authors
- Disclosure requirements for authors
- AI disclosure statement: standard format and placement
- AI in data analysis, statistical processing, and code generation
- AI-generated and AI-modified images and visual materials
- AI and bibliographic references
- AI and supplementary materials
- AI and medical writers
- AI use by peer reviewers
- AI use by editors and editorial board members
- AI in the editorial workflow: publisher and editorial office
- AI as subject of study in submitted manuscripts
- Identification of violations and editorial response
- Sanctions
- Reporting channels for suspected AI-related misconduct
- Limits of AI detection tools
- Data protection and confidentiality in AI use (master clause)
- Scope of application and transitional provisions
- Annual review of this Policy
- Cross-references within the Editorial Policy Framework
- Exemptions and clarifications
- Reference standards
- SCOPE AND APPLICABILITY
This Artificial Intelligence Policy establishes the rules governing the use of artificial intelligence tools in all phases of the research, writing, preparation, peer review, editorial management, and publication process of European Gynecology and Obstetrics (EGO).
The Policy applies to:
- authors of manuscripts submitted to the Journal, regardless of article type;
- peer reviewers assigned to any manuscript under consideration;
- members of the Editorial Board, Associate Editors, Section Editors, and the Editor-in-Chief;
- Guest Editors of special issues and supplements;
- Editorial Fellows and early-career mentors;
- the Editorial Office and operational staff;
- Edikta S.r.l. in its capacity as Publisher, for the use of automated tools in the editorial workflow.
For this Policy, "artificial intelligence tools" (hereinafter: AI tools) means any software, application, platform, or model that employs machine learning, large language models (LLMs), generative algorithms, or other AI-based techniques to generate, translate, edit, summarise, classify, or analyze text, images, data, code, or other content. The definition encompasses, by way of example and not exhaustively: generative text models (ChatGPT, Claude, Gemini, Copilot, DeepSeek, and equivalent systems), image generation models, AI-assisted coding tools, AI-assisted statistical analysis tools, and AI-powered literature review assistants, whether accessed via web interface, API, desktop application, or integrated into third-party software.
For this Policy, the "Editorial Direction" means the Editor-in-Chief acting together with the Publisher or the Editor-in-Chief acting alone where this Policy or the editorial framework of the Journal so provides. In matters of day-to-day editorial administration, the Editorial Direction may be represented by the designated Managing Editor.
This Policy applies to all manuscripts submitted to the Journal from the date of its publication. For manuscripts submitted before publication of this Policy, the provisions on disclosure and authorship apply retroactively to the extent compatible with the stage of the editorial process already reached. The Policy is read in conjunction with the Authorship Policy, the Peer Review Policy, the Publication Ethics Policy, the Editorial Policy Statement, and the Corrections and Retractions Policy as part of the integrated editorial policy framework of the Journal.
- GOVERNING PRINCIPLES
The Journal's approach to artificial intelligence is founded on 4 non-negotiable governing principles, consistent with the positions of COPE, ICMJE, WAME, EASE, and the STM Association Generative AI Guidelines (2024):
- Human authorship and responsibility. Only human beings can be authors of scientific work. Authors bear full, non-delegable responsibility for the accuracy, integrity, and originality of all content in a submitted and published manuscript, including any content generated, drafted, or edited with AI assistance. This responsibility cannot be transferred to or shared with any AI tool.
- Transparency and disclosure. Any use of AI tools in the preparation of a manuscript or in the peer-review process must be disclosed transparently, with sufficient specificity to enable editors, reviewers, and readers to assess the nature and extent of AI assistance.
- Scientific integrity. Authors must independently verify all content produced with AI assistance - including factual claims, bibliographic references, statistical outputs, and image content - before submission. AI tools may produce outputs that are incorrect, incomplete, fabricated, or biased. Verification is a non-delegable author responsibility.
- Protection of confidentiality (master clause). The unpublished manuscript, associated data, peer review reports, editorial communications, and all related materials constitute confidential information. No person involved in the editorial process - whether author, reviewer, editor, editorial staff member, or Publisher representative - may upload, transmit, copy, paste, or otherwise expose any such confidential information to any external AI platform or service, regardless of the data retention or privacy policies declared by the AI provider, and regardless of any prior de-identification attempt. This prohibition encompasses all generative AI platforms accessible via a web interface, an API, a desktop application, or an integration with third-party software. The full data protection and confidentiality framework applicable to the use of AI in the editorial process is set out in section 20 of this Policy.
- AI TOOLS AND AUTHORSHIP
AI tools - including generative AI systems, large language models, and any equivalent technology - cannot be listed as authors or co-authors of manuscripts submitted to the Journal. This prohibition is absolute and admits no exceptions.
Authorship requires: (a) substantial contributions to the conception, design, acquisition, analysis, or interpretation of data; (b) drafting or critically revising the manuscript for important intellectual content; (c) final approval of the version to be published; (d) agreement to be accountable for all aspects of the work. AI tools are incapable of fulfilling any of these criteria. In particular, AI tools cannot: assume legal responsibility for the accuracy and integrity of the work; give or withhold consent to publication; respond to post-publication queries or investigations; declare the absence or presence of conflicts of interest; or enter into publishing agreements. Any manuscript listing an AI tool in the author position will be desk-rejected without review.
AI tools may not be cited as sources in the reference list. AI-generated outputs do not constitute peer-reviewed, authoritative scientific sources. Authors who have used an AI tool to obtain information, summaries, or analyses must independently verify the accuracy of such information against primary, verifiable sources and cite those primary sources in the manuscript. Citing AI-generated content as a primary source is treated as a form of misrepresentation and may result in the measures set out in section 17 of this Policy.
- PERMITTED AND PROHIBITED USES OF AI BY AUTHORS
Substantive threshold for AI-assisted writing. AI tools may assist in phrasing, language, and expression, but may not generate original scientific argumentation, interpretation of results, or novel intellectual content. AI-assisted drafting is permitted only where the intellectual structure, the argumentation, and the final wording are substantially determined and controlled by the human authors. Submission of minimally edited AI-generated text, where the substantive intellectual content originates from the AI tool rather than the authors, constitutes prohibited ghostwriting and is subject to the measures set out in section 17 of this Policy.
The following uses of AI tools by authors are permitted, subject to compliance with the disclosure requirements set out in sections 5 and 6 of this Policy:
- Language editing and proofreading (spelling, grammar, syntax, readability), provided that the intellectual content and scientific meaning of the text are not altered;
- Translation of the manuscript or sections thereof between languages, with subsequent verification by the authors that the scientific meaning has been preserved;
- Reformatting of references, bibliographic metadata, or structured data, with independent verification of the output;
- Generation of structured outlines or organizational frameworks for the manuscript, provided the intellectual content is entirely authored by the human authors;
- Literature search support (identification of potentially relevant articles), provided that authors independently verify the existence, accuracy, and relevance of every retrieved reference;
- Summarization of author-owned datasets or texts for the exclusive purpose of facilitating the authors' own drafting process, provided the final text is independently authored and the substantive threshold set out above is respected;
- Statistical analysis and data processing using AI-based tools, subject to the additional requirements set out in section 7 of this Policy;
- Code generation for data analysis, subject to the additional requirements set out in section 7 of this Policy;
- Use of AI tools to assist in drafting the cover letter, abstract, keywords, or suggested reviewer information submitted at the time of manuscript submission, subject to disclosure in accordance with section 5 of this Policy.
The following uses of AI tools by authors are prohibited:
- Uploading confidential manuscript materials to external AI platforms: in accordance with the master clause set out in section 2, principle 4, and with section 20 of this Policy.
- Using AI tools to fabricate, hallucinate, or generate fictitious data, results, references, or quotations;
- Using AI tools to generate or substantially modify images, figures, photographs, radiological images, histological images, or other visual scientific content, except as expressly permitted under section 8 of this Policy;
- Relying on AI-generated bibliographic references without independent verification of each source;
- AI-assisted writing where the intellectual structure, argumentation, and final wording are not substantially determined by the human authors (ghostwriting), as defined in the substantive threshold set out above in this section;
- Using AI tools to circumvent plagiarism detection, to restate text from third-party works in ways designed to avoid detection, or to conceal duplicate publication;
- Embedding hidden text, invisible characters, zero-width characters, micro-font text, white-on-white text, metadata fields, or any other form of concealed content in the manuscript, supplementary materials, or any submission file, to provide instructions to AI tools used in the editorial workflow (prompt injection attacks). Such conduct constitutes deliberate manipulation of the editorial process and is treated as serious research misconduct, subject to the measures set out in section 17 of this Policy.
The following actions do not require disclosure under this Policy:
- Basic spell-checking and standard grammar correction functions integrated into standard word processors (e.g., Microsoft Word spell check, standard autocorrect);
- Standard search engine queries and database searches (PubMed, Embase, Cochrane, Scopus, Web of Science) that do not involve LLM-based generative summarization;
- Use of reference management software (Mendeley, Zotero, EndNote) for citation formatting, provided no generative AI function is used.
- DISCLOSURE REQUIREMENTS FOR AUTHORS
Authors must disclose any use of AI tools that falls within the categories requiring disclosure as identified in section 4 of this Policy. The disclosure obligation arises at the moment of submission and applies to all versions of the manuscript, including revised versions submitted in response to peer review.
The disclosure must contain, at minimum, the following elements:
- the name of the AI tool or platform used;
- the version of the tool or, where no version number is available, the date of access;
- the specific purpose for which the tool was used (e.g., language editing, statistical analysis, literature search support, translation, code generation, cover letter drafting, abstract drafting);
- the sections or parts of the manuscript affected by AI assistance;
- a statement confirming that all AI-generated or AI-assisted content has been independently reviewed, verified, and approved by the named authors, and that the substantive threshold set out in section 4 has been respected.
A single disclosure statement covering all AI use in the manuscript is acceptable. Where the manuscript includes multiple distinct AI tools used for different purposes, each tool must be identified separately.
Where AI tools have been used in the preparation of the cover letter, abstract, keywords, or suggested reviewer information, this must be disclosed in the cover letter itself, in the designated submission system field, or in the AI disclosure statement of the manuscript, as appropriate.
Where a manuscript has previously been deposited as a preprint and AI tools were used in the preparation of the preprint, the AI disclosure submitted with the journal submission must cover the entire work, including any AI use in the preprint phase. Prior disclosure in the preprint does not discharge the author's disclosure obligations under this Policy. The rules of this Policy apply to the submitted manuscript in its entirety, regardless of the version history of the work.
Failure to disclose AI use that is subsequently identified constitutes misrepresentation and is subject to the measures set out in section 17 of this Policy. Whether the failure to disclose constitutes a minor omission or a material breach will be assessed based on the nature of the undisclosed AI use, its impact on the integrity of the manuscript, and whether the non-disclosure appears to have been deliberate.
- AI DISCLOSURE STATEMENT: STANDARD FORMAT AND PLACEMENT
The AI disclosure statement must be included in the body of the manuscript as a distinct, labeled section. Depending on the nature of the AI use, placement follows the rules below:
- AI use for language editing, proofreading, translation, or reformatting: the disclosure statement is placed in the Acknowledgments section, clearly separated from acknowledgments of personal or institutional support.
- AI use for data analysis, statistical processing, code generation, or literature search: the disclosure statement is placed in the Methods section, immediately adjacent to the description of the methodology to which AI contributed.
- AI use spanning multiple categories: a combined AI disclosure statement may be placed at the end of the Methods section, with a cross-reference in the Acknowledgments section.
The following standard templates are provided for authors' use:
Template A - Language editing: "[Tool name] ([version/access date]) was used to improve the language and readability of sections [specify] of this manuscript. The authors reviewed and approved all final text. The AI tool was not used to generate scientific content, analyze data, or produce conclusions."
Template B - Data analysis or code: "[Tool name] ([version/access date]) was used to [describe specific use] in the Methods section. All outputs were independently verified by [author name(s)]. The authors determined the final results and their interpretation."
Template C - No AI use: "The authors declare that no artificial intelligence tools were used in the preparation of this manuscript."
These templates may be adapted to the specific circumstances of each manuscript. Authors are responsible for ensuring that the disclosure is accurate, complete, and sufficiently specific.
- AI IN DATA ANALYSIS, STATISTICAL PROCESSING, AND CODE GENERATION
The use of AI tools for data analysis, statistical processing, or code generation is permitted subject to the following conditions:
- Authors must disclose the AI tool used, its version or access date, the nature of the analysis or code generation task performed, and the dataset or input to which the tool was applied, in the Methods section of the manuscript.
- All AI-generated analytical outputs - including statistical results, summary tables, visualizations, and interpretations - must be independently verified by at least one author with the relevant methodological or statistical competence before inclusion in the manuscript.
- All AI-generated code used in the analysis must be reviewed, tested, and validated by a human author before use. The code must be made available as supplementary material or in a designated open repository, in accordance with the Journal's Data Availability and Sharing Policy.
- Authors may not delegate the interpretation of results to an AI tool. The scientific interpretation of findings, including the discussion of their significance, limitations, and implications, must be authored entirely by the human authors.
- When AI tools are used in automated or high-throughput data-processing pipelines, the authors must describe the pipeline architecture and its AI components in sufficient detail to enable independent replication.
The use of AI for statistical analysis does not replace the requirement to comply with applicable reporting guidelines (CONSORT, STROBE, PRISMA, CARE, STARD, ARRIVE, SAGER, or equivalent), as set out in the Journal's Statistical Reporting and Reproducibility Policy (SRRP).
- AI-GENERATED AND AI-MODIFIED IMAGES AND VISUAL MATERIALS
The following rules apply to all images, figures, graphs, radiological images, histological images, photographs, diagrams, and other visual materials included in a manuscript submitted to the Journal.
Definitions. For this section, "standard image adjustments" means operations that do not alter the scientific content of an image and that are applied uniformly to the entire image: format conversion, resizing, uniform brightness and contrast adjustment, and uniform color balance correction. "Non-standard image adjustments" means any operation that selectively alters part of an image, introduces or removes content, or applies corrections non-uniformly, including: selective brightness or contrast manipulation, background removal or replacement, object insertion, cropping that omits scientifically relevant content, noise reduction that alters cellular or morphological detail, and resolution enhancement that introduces non-existent detail. This definition is consistent with the image integrity standards of COPE, the Journal of Cell Biology, and the STM image integrity guidelines.
- Prohibition on AI-generated clinical and scientific images. The generation of clinical images, histological images, radiological images, photographic documentation of patients or specimens, or any other image that purports to represent actual experimental, clinical, or observational data, using AI generative tools, is strictly prohibited. Such images are considered fabricated scientific data and will result in the measures set out in section 17 of this Policy being applied.
- Prohibition on undisclosed non-standard AI image adjustments. Non-standard adjustments applied using AI tools must be disclosed. Undisclosed non-standard AI image modification is treated as image manipulation in accordance with the Corrections and Retractions Policy of the Journal and with the definition set out above.
- Standard AI-assisted image processing: no disclosure required. Standard image adjustments (as defined above) do not require AI-specific disclosure, provided they comply with the Journal's general image integrity standards.
- AI-generated illustrative and schematic content. AI tools may be used to generate purely illustrative, schematic, or conceptual diagrams (e.g., workflow diagrams, anatomical schematics, visual abstracts) that do not purport to represent real experimental or clinical data, provided that such use is explicitly disclosed in the figure legend and in the AI disclosure statement.
- Figure legend disclosure. Where AI tools have been used in the non-standard modification or generation of any image included in the manuscript, the figure legend must include an explicit statement identifying the AI tool used and describing the modification performed.
The Journal reserves the right to request original, unmodified image files at any stage of the editorial process, including after acceptance, for image integrity verification. Authors must retain the original image files for a minimum of 5 years following publication.
- AI AND BIBLIOGRAPHIC REFERENCES
Authors who use AI tools in any part of the literature review, reference identification, or citation process are subject to the following requirements:
- The authors must independently verify every bibliographic reference cited in the manuscript before submission. Verification means: (a) confirming that the cited work exists and has been published or deposited as described; (b) confirming that the cited work contains the content attributed to it in the manuscript; (c) confirming that the DOI or other persistent identifier is accurate and resolves to the correct source.
- References that cannot be independently verified must not be included in the manuscript. AI-generated references that do not correspond to existing published works ("hallucinated" references) constitute a form of fabrication and are subject to the measures set out in section 17 of this Policy.
- AI tools may not be cited as primary sources in the reference list. If an AI-generated output has been used to obtain information subsequently verified against a primary source, the primary source must be cited.
- Authors must disclose the use of AI tools for literature search or reference generation in the AI disclosure statement, as set out in section 5 of this Policy.
- AI AND SUPPLEMENTARY MATERIALS
The disclosure and integrity requirements set out in this Policy apply in full to all supplementary materials submitted with or linked to a manuscript, including supplementary data files, supplementary figures and tables, supplementary methods sections, and analysis code or scripts.
Where AI tools have been used in the preparation of supplementary materials, this must be disclosed in the AI disclosure statement of the main manuscript, with specific reference to the supplementary materials affected. The confidentiality requirements of section 20 apply in full to supplementary data.
- AI AND MEDICAL WRITERS
Where a professional medical writer, editorial assistant, or communication agency has contributed to the preparation of the manuscript, any use of AI tools by the medical writer in the course of that contribution must be disclosed. Responsibility for the completeness and accuracy of the AI disclosure statement rests with the named authors, who are obliged to obtain from the medical writer the information necessary for full disclosure.
In accordance with section 20 of the Authorship Policy of the Journal, where a medical writer uses AI tools in assisting, this must be reported as part of the AI use disclosure statement. The nature of the AI assistance provided by the medical writer must be described with the same specificity required for any other AI use (tool name, version, purpose, sections affected).
The use of AI tools by a medical writer does not reduce or modify the responsibility of the named authors for the accuracy, integrity, and scientific content of the manuscript.
- AI USE BY PEER REVIEWERS
The peer review process depends on the independent, expert judgment of qualified reviewers. The confidentiality and integrity of the process must be preserved.
Prohibition on uploading manuscript materials. In accordance with section 2, principle 4 (master clause) and section 20 of this Policy, reviewers may not upload, transmit, copy, paste, or otherwise expose any part of the manuscript under review, its supplementary materials, the review report being prepared, or any related editorial communication, to any external AI platform. This prohibition is absolute and applies regardless of any attempt to de-identify the manuscript.
Rejection of the partial de-identification approach. Some publishing organizations permit reviewers to use AI tools provided that identifying elements are removed before upload. The Journal does not adopt this approach, for the reasons set out in section 20: complete and reliable de-identification cannot be guaranteed, and any upload of manuscript content to an external system compromises the confidentiality of unpublished work and the double-anonymized review process.
Limited permitted use. Reviewers may use AI tools exclusively for: (a) improving the language, spelling, or grammar of their own written review report, provided no part of the manuscript content is included in the AI query; (b) checking the mathematical or formal structure of their own review text, using secure, locally-run tools that do not transmit data to external servers. The reviewer must disclose any such use in the confidential comments field of the review report.
Formal authorization procedure for specific tools. In particular and duly reasoned cases, a reviewer may submit a written request to the Editorial Direction for authorization to use a specific AI tool for a defined, limited purpose in the review process. The request must specify: (a) the name and version of the tool; (b) the intended purpose of use; (c) the technical modality (local processing versus cloud-based); (d) the data retention and privacy policy of the tool provider; (e) the precise scope of application within the review process. The Editorial Direction responds in writing. In the absence of explicit written authorization, any use beyond the limited permitted uses above is prohibited. Each authorization granted is documented in the editorial record of the manuscript concerned.
Disclosure by reviewers. Where a reviewer has made any permitted use of AI tools or has received written authorization for a specific tool, this must be declared in the "Comments confidential to the Editorial Direction" field of the review report, specifying: the tool used, its version or access date, and the specific limited purpose for which it was employed. Reviewers who have not used any AI tools must confirm this in the same field.
Consequences of breach. Reviewers who violate the prohibition on uploading manuscript materials, or who submit AI-generated review reports without disclosure, will be removed from the Journal's reviewer database. The Editorial Direction will inform the Editor-in-Chief and will determine whether the affected manuscript requires re-review. In cases of evident breach, the matter will be referred to the reviewer's institution in accordance with COPE guidance.
- AI USE BY EDITORS AND EDITORIAL BOARD MEMBERS
Editors, Associate Editors, Section Editors, and Editorial Board members are subject to the same confidentiality and disclosure obligations as peer reviewers with respect to unpublished manuscripts.
Prohibition on upload. In accordance with section 2, principle 4 (master clause) and section 20 of this Policy, no editor or Editorial Board member may upload any part of an unpublished manuscript or associated materials to an external AI platform for the purpose of producing or informing an editorial decision.
- Editors may use AI tools for administrative or language-related tasks that do not involve exposure of manuscript content, provided such use does not substitute for independent editorial judgment.
- Editorial Board members who have used AI tools in any capacity in connection with an editorial function must disclose this to the Editor-in-Chief promptly.
- The Editorial Office’s use of AI tools for editorial screening is governed by section 14 of this Policy and is subject to mandatory human oversight.
- Where an editor, Associate Editor, Section Editor, or Editorial Board member holds a financial interest in, is employed by, or has an ongoing consultancy relationship with a company that develops or commercializes an AI tool used in the editorial workflow of the Journal, this constitutes a conflict of interest that must be disclosed to the Editor-in-Chief and managed in accordance with the Editorial Board Charter and the Conflict of Interest Policy of the Journal. The editor concerned must recuse from any editorial decision in which the relevant AI tool has been applied.
- AI IN THE EDITORIAL WORKFLOW: PUBLISHER AND EDITORIAL OFFICE
Edikta S.r.l., in its capacity as Publisher, and the Editorial Office may employ AI-based or automated tools in the editorial workflow for quality assurance, research integrity, and operational efficiency, governed by the following principles:
- Permitted uses. Automated tools may be employed for: text similarity and plagiarism detection (including AI-content detection tools); image integrity and duplication checks; technical formatting verification; metadata validation; reference verification; reviewer suggestion support; and routine administrative tasks.
- Human oversight mandatory. All outputs of automated editorial tools are subject to mandatory human review and verification by editorial staff before any action is taken. No editorial decision - including desk rejection, acceptance, retraction, or expression of concern - may be made on the sole basis of an automated tool output.
- Transparency toward authors. The Journal will publicly communicate, in its instructions for authors and on its website, the categories of automated tools used in the editorial workflow. Authors will be notified when an automated tool has generated an alert that has influenced an editorial decision concerning their manuscript. They will be allowed to respond before any adverse decision is made.
- AI detection tools: limitations. The Journal acknowledges the documented limitations of AI content detection tools, including significant false-positive rates and systematic biases affecting non-native English speakers and authors from certain geographic regions. Detection tool outputs are treated as indicators requiring human verification, not as conclusive evidence of AI misconduct. The provisions of section 19 of this Policy apply.
- Confidentiality obligations. In accordance with section 2, principle 4 (master clause) and section 20 of this Policy, no unpublished manuscript or associated materials may be uploaded to an external generative AI platform by Editorial Office staff or Publisher representatives.
- AI AS SUBJECT OF STUDY IN SUBMITTED MANUSCRIPTS
European Gynecology and Obstetrics (EGO) publishes research in obstetrics and gynecology in which artificial intelligence is also a subject of scientific study - for example, AI diagnostic tools, AI-assisted imaging, AI-based clinical decision support systems, or AI in population health research. This section distinguishes between AI as a writing and analysis tool (governed by the preceding sections of this Policy) and AI as a clinical or research intervention studied in the manuscript.
Where AI is the subject of study or the primary intervention in a submitted manuscript:
- The manuscript is assessed for scientific merit and methodological rigor in the same manner as any other submission, in accordance with the Journal's Peer Review Policy.
- Clinical trials evaluating AI interventions initiated on or after 1 January 2025 must be prospectively registered in a WHO-compliant trial registry (ClinicalTrials.gov, EU Clinical Trials Register, or WHO ICTRP) before the start of patient enrolment. Proof of registration must be provided at submission.
- Studies involving AI algorithms applied to patient data must demonstrate compliance with applicable data protection requirements, including the Italian Personal Data Protection Code (Legislative Decree 196/2003, as amended by Legislative Decree 101/2018) and the General Data Protection Regulation (EU) 2016/679. They must include a statement of ethics committee approval.
- Authors reporting on AI diagnostic or clinical decision tools are encouraged to follow the TRIPOD+AI reporting guideline (Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis - Artificial Intelligence extension) where applicable.
- The use of AI writing or analysis tools by the authors of such manuscripts remains subject to all requirements of this Policy.
- IDENTIFICATION OF VIOLATIONS AND EDITORIAL RESPONSE
The Journal employs the following mechanisms for the identification of potential violations of this Policy:
- Pre-submission screening: AI-content detection tools are applied to all submitted manuscripts as part of the technical quality check, subject to the limitations described in section 19.
- Image integrity check: automated image analysis tools are applied to all submitted images to detect potential manipulation or generation.
- Reference verification: editorial staff verifies a sample of references for existence and accuracy; AI-hallucinated references are reported as a red flag indicator.
- Hidden content screening: submitted manuscript files are inspected for the presence of hidden text, invisible characters, zero-width characters, micro-font text, white-on-white text, and metadata-embedded instructions that may constitute prompt injection attempts, as described in section 4 of this Policy.
- Peer review: reviewers are invited to report any indicator of undisclosed AI use in the comments confidential to the Editorial Direction.
- Post-publication monitoring: readers, authors, and third parties may report suspected violations in accordance with section 18 of this Policy.
Where an indicator of a potential violation is identified, the Editorial Direction proceeds as follows:
Step 1 - Internal verification. The Editorial Direction reviews the indicator with human oversight and determines whether a prima facie case for investigation exists.
Step 2 - Author notification. If a prima facie case is identified, the corresponding author is notified in writing and given a minimum of 10 working days to provide a response, explanation, or clarification.
Step 3 - Assessment. The Editorial Direction, in consultation with the Editor-in-Chief, assesses the author's response. Where the matter involves complex technical or scientific judgments, the advice of a subject-matter expert or statistical reviewer may be sought.
Step 4 - Decision and action. Based on the assessment, the Editorial Direction determines the appropriate response under section 17 of this Policy. All decisions are documented in the editorial management system.
At all stages, the Journal follows the applicable COPE flowcharts for cases of suspected misconduct.
- SANCTIONS
The following responses are available to the Journal in cases of confirmed violation of this Policy, applied proportionately to the nature, severity, and intentionality of the violation:
Pre-publication measures:
- Desk rejection of the manuscript without review, where the violation is identified before acceptance (e.g., AI tool listed as author; manifest AI-generated fabrication of data or images; confirmed prompt injection attempt).
- Rejection following peer review, where the violation is identified during or after review but before acceptance.
- Mandatory revision and supplementation of the disclosure statement, where the violation consists of an incomplete or inaccurate disclosure that can be corrected without affecting the scientific integrity of the manuscript.
Post-publication measures:
- Correction notice, where the violation is identified after publication and consists of a disclosure omission that does not affect the scientific findings.
- Expression of concern, where the violation raises concerns about the integrity of the findings that cannot be resolved promptly.
- Retraction, where the violation is material - in particular where AI use has resulted in the fabrication or falsification of data, images, or results, or where an undisclosed AI tool has produced content that renders the published findings unreliable.
Administrative measures:
- Removal of a reviewer from the Journal's reviewer database for breach of confidentiality or submission of undisclosed AI-generated review reports.
- Notification of the author's or reviewer's institutional affiliation, in cases of serious or intentional violation, in accordance with COPE guidance on institutional notification.
- Temporary or permanent exclusion of an author from submitting to the Journal, where a pattern of AI-related misconduct is established.
In all cases, the distinction between inadvertent omission and deliberate misrepresentation is taken into account in determining the appropriate response. Authors are allowed to respond at the investigation stage as set out in section 16.
- REPORTING CHANNELS FOR SUSPECTED AI-RELATED MISCONDUCT
Any author, reviewer, reader, editor, or third party who has reasonable grounds to suspect a violation of this Policy in a submitted or published manuscript may report their concern to the Editorial Direction of the Journal through the dedicated contact channel published on the Journal's official website.
Reports must contain: the title and, where available, the DOI or manuscript identifier of the manuscript concerned; a description of the suspected violation; and, where applicable, supporting evidence. Reports may be submitted anonymously. The identity of reporting parties who are not anonymous will be kept confidential throughout the investigation.
The Editorial Direction will acknowledge receipt of all reports within 5 working days. It will provide a substantive response regarding the outcome of the investigation within the timelines established by the applicable COPE flowcharts. Reporting parties who have acted in good faith will not be subject to adverse consequences as a result of their report, regardless of the outcome of the investigation.
- LIMITS OF AI DETECTION TOOLS
The Journal acknowledges that automated AI-content detection tools currently available present significant limitations that prevent their use as sole determinants of editorial decisions:
- False positive rates: AI detection tools may flag as AI-generated content that was entirely human-written, with documented higher error rates for non-native English speakers, authors from certain geographic regions, and highly technical or formulaic text;
- False negative rates: AI detection tools may fail to detect AI-generated content that has been subsequently edited or paraphrased by human authors;
- Absence of validated, peer-reviewed standards: most commercially available AI detection tools have not been independently validated through peer-reviewed scientific studies;
- Absence of an international consensus threshold: there is no internationally agreed threshold above which AI detection scores constitute conclusive evidence of AI use.
In view of these limitations, the Journal adopts the following operational rules:
- AI detection tool outputs are used as indicators requiring human verification, not as conclusive evidence;
- No editorial decision adverse to an author may be taken solely based on an AI detection tool output;
- Authors flagged by an AI detection tool will be notified and allowed to provide clarification and, where applicable, to supplement their AI disclosure statement;
- The Journal will periodically review the AI detection tools employed in its editorial workflow to assess their performance and to update its practice in line with evolving scientific evidence.
- DATA PROTECTION AND CONFIDENTIALITY IN AI USE (MASTER CLAUSE)
This section constitutes the master clause on confidentiality in AI use, to which sections 2 (principle 4), 4, 12, 13, and 14 of this Policy refer. The full prohibition, its rationale, and the applicable data protection obligations are set out here.
Confidentiality of unpublished materials. The unpublished manuscript, associated data, peer review reports, editorial communications, supplementary materials, and all related content constitute confidential information from the moment of submission until publication. No person involved in the editorial process - whether author, reviewer, editor, editorial staff member, or Publisher representative - may upload, transmit, copy, paste, or otherwise expose any such confidential information to any external AI platform or service. This prohibition: (a) applies regardless of the data retention or privacy policies declared by the AI provider; (b) applies regardless of any prior or attempted de-identification of the manuscript or its contents; (c) encompasses all generative AI platforms accessible via web interface, API, desktop application, or integrated into third-party software; (d) encompasses both text and non-text content (images, datasets, code, audio, video).
Rationale for the total prohibition. The prohibition is total and does not admit of partial exceptions for the following reasons: (a) it is impossible to verify that content submitted to external AI platforms is not retained, used for model training, or exposed to third parties; (b) complete and reliable de-identification of a scientific manuscript - which may contain topic-specific terminology, institutional identifiers, and referenced works - cannot be guaranteed by any party acting unilaterally; (c) even a partially de-identified manuscript constitutes a breach of the authors' right to confidentiality over their unpublished intellectual property.
- Where AI tools are employed by Edikta S.r.l. in the editorial workflow (as described in section 14), such tools must operate based on a data processing agreement that ensures: no retention of manuscript data beyond the purpose of the specific processing task; no use of manuscript data for model training; compliance with the General Data Protection Regulation (EU) 2016/679 and applicable Italian data protection law.
- Authors submitting manuscripts that include personal data (patient data, clinical data, genetic data, or other special categories of data under GDPR Article 9) must not use external AI tools to process or analyse such data unless: (a) they have verified that the AI tool operates under a data processing agreement compliant with GDPR; (b) the use has been approved by the relevant ethics committee or institutional data protection officer; and (c) the use is disclosed in the AI disclosure statement.
- Given the subject matter of the Journal - obstetrics, gynecology, reproductive medicine, and perinatal care - manuscripts frequently contain data belonging to special categories under GDPR Article 9, including: genetic data from prenatal or preimplantation testing; data concerning reproductive health, pregnancy history, abortion, and fertility treatment; data relating to gender identity in obstetric and gynecological contexts; clinical data on minors (newborns, neonates, and foetuses). Authors must exercise heightened caution before using any AI tool to process, analyze, or generate text from such data. In cases of doubt, processing of GDPR Article 9 data with AI tools is not permitted without prior authorization from the competent ethics committee or data protection officer.
- The use of AI tools to process research participants' personal data is subject to the additional requirements of the Journal's Research Ethics Policy.
- SCOPE OF APPLICATION AND TRANSITIONAL PROVISIONS
This Policy applies to all manuscripts submitted to the Journal from the date of its first publication on the Journal's official website.
For manuscripts submitted before the date of publication of this Policy and still under review at the date of publication, the following transitional provisions apply:
- Authors of manuscripts under review at the date of publication of this Policy will be notified by the Editorial Direction and invited to submit an AI disclosure statement within 14 calendar days.
- Manuscripts accepted for publication before the date of publication of this Policy are not subject to retroactive application, except where a specific violation (e.g., fabrication of data, undisclosed AI authorship) is independently identified.
- ANNUAL REVIEW OF THIS POLICY
Given the rapid pace of development of artificial intelligence technologies and the ongoing evolution of international standards and publisher practices in this field, this Policy will be reviewed annually by the Editorial Direction and the Publisher.
The annual review will assess: (a) developments in the positions of COPE, ICMJE, WAME, EASE, STM, and other relevant international bodies; (b) developments in AI technology that may affect the scope or application of the Policy; (c) the experience of the Journal in implementing the Policy, including any recurring issues or ambiguities; (d) developments in the scientific literature on AI detection tools and their reliability.
Material revisions to this Policy will be communicated to authors, reviewers, and Editorial Board members via the Journal's official communication channels. A versioned record of all editions of this Policy will be maintained and publicly accessible on the Journal's website.
- CROSS-REFERENCES WITHIN THE EDITORIAL POLICY FRAMEWORK
This Policy is part of the integrated editorial policy framework of European Gynecology and Obstetrics (EGO). The following cross-references apply:
- Authorship Policy, Section 19: establishes the 3 fundamental principles on AI and authorship that this Policy develops operationally;
- Authorship Policy, Section 20: governs medical writers' use of AI tools;
- Peer Review Policy, Section 13: establishes the general prohibition on uploading manuscript materials to external AI platforms and the formal authorization procedure for specific tools;
- Editorial Policy Statement, Section 18: establishes the founding editorial principle on AI use and transparency, of which this Policy is the operational implementation;
- Corrections and Retractions Policy: governs post-publication measures in cases of AI-related misconduct;
- Publication Ethics Policy: establishes the overall ethical framework within which this Policy operates;
- Editorial Board Charter and Conflict of Interest Policy: govern the management of conflicts of interest arising from editors' affiliations with AI companies;
- Research Ethics Policy: governs clinical trials involving AI interventions and the processing of patient data with AI tools;
- Statistical Reporting and Reproducibility Policy: governs reporting of AI-assisted statistical analyses;
- Data Availability and Sharing Policy: governs the availability of AI-generated code and datasets in open repositories;
- Preprint Policy: governs the continuity of AI disclosure obligations from preprint deposit to journal submission.
- EXEMPTIONS AND CLARIFICATIONS
The following situations are clarified for the avoidance of doubt:
- Standard spell-checking, grammar-checking, and autocorrect functions integrated in word processors (e.g., Microsoft Word, Google Docs) are not subject to disclosure requirements under this Policy, unless they are powered by a generative AI component that the user has actively engaged to produce or substantially modify text.
- Reference management software (Zotero, Mendeley, EndNote) used for citation formatting and reference organization does not require AI disclosure, unless the user has actively employed an AI-powered feature of the software.
- Standard statistical software (R, SPSS, SAS, Stata, MATLAB) does not require AI disclosure, unless the user has employed AI-powered features of the software beyond standard algorithmic computation.
- Search engines and databases (PubMed, Embase, Cochrane, Scopus, Web of Science) accessed via their standard search interface do not require AI disclosure, unless the user has employed an integrated AI summarization or generation feature.
- Translation services used exclusively for non-scientific administrative correspondence are not subject to this Policy.
Authors who are uncertain whether a specific use of an AI tool requires disclosure are encouraged to contact the Editorial Direction for guidance before submission. The principle of the Journal is that, in cases of doubt, disclosure is preferable to non-disclosure.
- REFERENCE STANDARDS
This Policy is consistent with and draws upon the following international standards and guidance documents:
- Committee on Publication Ethics (COPE). COPE Position Statement: Authorship and AI tools. 2023 (updated 2024).
- Committee on Publication Ethics (COPE). Guidance on AI and peer review. 2024.
- International Committee of Medical Journal Editors (ICMJE). Recommendations for the Conduct, Reporting, Editing, and Publication of Scholarly Work in Medical Journals. Updated April 2025.
- World Association of Medical Editors (WAME). Chatbots, Generative AI, and Scholarly Manuscripts: WAME Recommendations. 2023.
- European Association of Science Editors (EASE). Recommendations on the Use of AI in Scholarly Communication. 2024.
- STM Association. Generative AI in Scholarly Communications: Ethical and Practical Guidelines. 2024.
- Association for Computing Machinery (ACM). Principles of Peer Review. 2023.
- Springer Nature. AI Principles and Policies for Authors, Editors and Peer Reviewers. 2024 (updated 2025).
- Elsevier. Generative AI Policies for Journals. 2024.
- Wiley. Best Practice Guidelines on Research Integrity and Publishing Ethics: Artificial Intelligence. 2024.
- TRIPOD+AI. Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis - Artificial Intelligence extension. BMJ 2024.
- Journal of Cell Biology / Rockefeller University Press. Image Integrity Standards. 2024.
- STM Association. Image Integrity Guidelines. 2024.
- Regulation (EU) 2016/679 (General Data Protection Regulation - GDPR).
- Italian Personal Data Protection Code (Legislative Decree 196/2003, as amended by Legislative Decree 101/2018).