AI and Generative Tools

1. Purpose and Scope

AMG Transcend Association recognizes that artificial intelligence and generative technologies may be used in research, manuscript preparation, data analysis, scholarly communication, and publishing workflows.

This policy establishes publisher-level principles for the responsible use of artificial intelligence, machine-learning systems, large language models, generative AI, and related AI-assisted tools across the journals published by AMG Transcend Association.

The policy applies, as relevant, to authors, reviewers, Editors-in-Chief, editors, Editorial Board members, and editorial or publishing personnel.

Individual journals may establish additional or more restrictive requirements appropriate to their disciplines, editorial processes, research methods, and types of published content.

2. Fundamental Principle of Human Responsibility

Artificial intelligence does not replace human scholarly or editorial responsibility.

Authors remain responsible for all content submitted or published under their names, including text, references, data representations, analyses, images, code, interpretations, conclusions, and other material produced, modified, or assisted by AI tools.

Reviewers remain responsible for the accuracy, substance, integrity, and recommendations contained in their review reports.

Editors remain responsible for editorial assessments, judgments, and decisions.

The use of an AI system does not transfer, reduce, or remove the responsibilities associated with authorship, peer review, editorial judgment, research integrity, confidentiality, publication ethics, or compliance with applicable policies.

Users must independently evaluate and verify AI-assisted outputs before relying on them in scholarly, scientific, or editorial work.

3. AI Systems Cannot Be Authors

Artificial-intelligence systems, chatbots, language models, generative systems, or other automated tools must not be listed as authors or co-authors.

Authorship requires human intellectual contribution, approval, responsibility, and accountability that an AI system cannot assume.

AI tools must not be presented in a manner that falsely attributes human authorship, responsibility, accountability, or independent scholarly judgment to the system.

Where an AI system or tool forms part of the research methodology, it may be appropriately identified as a methodological or software resource, with relevant technical information provided where necessary.

Where AI use requires disclosure, that use must be described transparently rather than represented through authorship.

4. Use of AI in Manuscript Preparation

AI-assisted tools may be used in manuscript preparation where their use is consistent with scholarly integrity, applicable journal requirements, and full human responsibility for the resulting content.

Permitted uses may include, where appropriate:

  • language and grammar assistance;
  • stylistic or readability improvement;
  • translation;
  • organizational or structural assistance;
  • summarization used as part of an author-controlled drafting process;
  • reference-management assistance;
  • coding or computational assistance;
  • other functions permitted by the relevant journal.

Authors must critically review, verify, and take responsibility for all AI-assisted material before submission.

AI-generated or AI-assisted content must not introduce:

  • fabricated facts;
  • unsupported claims;
  • nonexistent or inaccurate references;
  • false quotations;
  • inappropriate reuse of third-party material;
  • misleading interpretations;
  • fabricated data or results;
  • other material inaccuracies or misrepresentations.

The use of AI does not relieve authors of responsibility for originality, attribution, copyright compliance, confidentiality, research integrity, or the accuracy and appropriateness of references.

5. Disclosure of AI Use

Material use of generative AI or other AI-assisted technologies in the preparation of scholarly content must be disclosed.

A disclosure should identify, where relevant:

  • the tool, system, or model used;
  • the purpose for which it was used;
  • the nature or extent of its contribution.

Material use may include substantive AI assistance in drafting or restructuring scholarly content, interpretation, analysis, coding, generation or modification of research outputs, or other activities that materially contribute to the submitted work.

Routine tools whose functions are limited to spelling correction, grammar checking, reference formatting, basic word-processing functions, or similarly minor technical assistance do not normally require disclosure unless otherwise required by the relevant journal.

AI use that forms part of the research methodology, including AI-based analysis, modeling, classification, prediction, image analysis, computational processing, or other substantive research procedures, must be reported in the Methods section or another appropriate part of the manuscript with sufficient detail for scientific evaluation.

A general AI disclosure does not replace methodological reporting where AI forms part of the research itself.

Disclosure of permitted AI use does not by itself imply misconduct or diminish the scholarly value of a manuscript.

Individual journals may establish additional disclosure requirements appropriate to their disciplines and article types.

6. AI as Part of the Research Methodology

Artificial intelligence, machine learning, computational models, or related technologies may legitimately form part of research methodology.

Where such methods materially contribute to the research, authors must report them with sufficient transparency to permit appropriate scientific assessment and, where reasonably possible, reproduction or independent verification of the analytical approach.

Relevant information may include:

  • the system, model, algorithm, software, or platform used;
  • the scientific purpose of its use;
  • relevant versions or configurations;
  • data sources and preprocessing;
  • training, testing, and validation procedures, where applicable;
  • important parameters or settings;
  • evaluation methods and performance measures;
  • human oversight;
  • relevant limitations;
  • other information necessary to understand or reproduce the analytical approach.

Where proprietary or externally hosted systems prevent full disclosure of technical details, authors must still provide sufficient information to permit meaningful scientific evaluation.

AI-based research methods are subject to the same expectations of scientific rigor, methodological validity, data integrity, reproducibility, ethics, and transparent reporting as other research methods.

7. AI-Generated and Synthetic Data

AI or computational methods may be used to generate synthetic, simulated, augmented, or modeled data where such use is scientifically justified and appropriate to the research design.

Such data must be clearly identified as generated, synthetic, simulated, augmented, or modeled and must not be presented as directly observed experimental, participant, clinical, or other empirical data.

Authors must describe the method of generation, validation where applicable, and the role of generated data with sufficient transparency for readers and reviewers to understand their provenance and purpose.

AI must not be used to fabricate research observations, experimental results, participant data, measurements, clinical information, or other evidence and present them as genuine empirical findings.

8. Data, Images, Figures, and Other Research Outputs

AI-assisted processing of data, images, figures, audio, video, spectra, graphical outputs, or other research materials must not misrepresent the underlying evidence.

Authors must distinguish legitimate scientific processing, enhancement, transformation, or analysis from alterations that create a false or misleading representation of research findings.

Where AI-assisted image, signal, data, or other processing materially affects the reported research, the relevant methodology must be described transparently.

AI tools must not be used to:

  • fabricate experimental images or data;
  • introduce or remove scientifically meaningful features without legitimate methodological justification;
  • selectively alter evidence in a misleading manner;
  • conceal manipulation;
  • create misleading representations of observations;
  • falsify the provenance of research materials.

Applicable requirements in the Data Integrity and Reproducibility, Image Manipulation, and Publication Ethics and Research Integrity policies remain in force.

9. References and Source Verification

AI systems may generate inaccurate, incomplete, inappropriate, misattributed, or nonexistent references.

Authors are responsible for verifying every reference included in a manuscript, regardless of how the reference was identified, suggested, or generated.

Authors must ensure that:

  • cited works exist;
  • bibliographic information is accurate;
  • references support the statements or context for which they are cited;
  • sources have been consulted sufficiently to support their use;
  • quotations and attributions are accurate;
  • citations are not fabricated or included merely because they were suggested by an AI system.

AI-generated output must not be treated as an authoritative substitute for the underlying scholarly, methodological, legal, ethical, or primary source.

10. Confidentiality and Sensitive Information

AI tools must not be used in ways that compromise confidentiality, privacy, intellectual property, research-participant protections, or the security of unpublished or restricted information.

Particular care is required when AI systems are used in connection with:

  • confidential research data;
  • personal or identifiable participant information;
  • sensitive datasets;
  • unpublished research material;
  • proprietary information;
  • information subject to contractual restrictions;
  • information protected by ethical, legal, regulatory, or institutional obligations.

Use of an external AI service must be compatible with applicable consent, ethics approval, privacy and data-protection requirements, contractual obligations, confidentiality duties, and intellectual-property rights.

Information must not be uploaded to or processed through an AI service where doing so would expose it to unauthorized retention, reuse, model training, disclosure, transfer, or third-party access contrary to applicable requirements.

11. Use of AI by Peer Reviewers

Peer review is a confidential process based on the reviewer’s independent scientific or scholarly expertise and judgment.

Reviewers must not upload submitted manuscripts, figures, supplementary files, confidential data, or substantive unpublished manuscript content to external AI systems or services unless:

  • use of the particular system has been explicitly permitted by the journal; and
  • applicable confidentiality, privacy, intellectual-property, security, and data-protection requirements are adequately protected.

AI tools must not replace the reviewer’s own critical scientific assessment or independent judgment.

Where expressly permitted by the relevant journal and compatible with confidentiality requirements, reviewers may use limited AI-assisted tools for functions such as language, formatting, or organizational assistance.

The reviewer remains fully responsible for the accuracy, substance, tone, confidentiality, and recommendation contained in the review.

Any material AI use in peer review must be disclosed where required by the relevant journal.

12. Use of AI by Editors and Editorial Personnel

Editorial decisions must remain under accountable human authority.

AI tools may support appropriate administrative, technical, analytical, or workflow functions, such as:

  • organization of editorial information;
  • technical compliance checks;
  • workflow assistance;
  • administrative communications;
  • similarity or integrity screening;
  • identification of potential issues requiring human review;
  • other non-decisional support functions.

AI-generated assessments, rankings, scores, classifications, or recommendations must not independently determine whether a manuscript is accepted, rejected, sent for peer review, assigned to a particular editor or reviewer, or otherwise evaluated.

Editors must independently assess relevant information and remain accountable for all editorial decisions.

Unpublished manuscripts, reviewer reports, confidential correspondence, and other protected editorial materials must not be submitted to external AI systems unless use of the system is authorized and applicable confidentiality, privacy, intellectual-property, security, and data-protection requirements are adequately protected.

13. AI Detection and Automated Screening

Automated tools may assist editors in identifying potential issues requiring further examination, but their outputs must not be treated as conclusive evidence of misconduct, inappropriate AI use, or another policy violation.

This principle applies to tools intended to detect or assess:

  • AI-generated text;
  • plagiarism or textual similarity;
  • image manipulation;
  • fabricated or unusual references;
  • statistical anomalies;
  • data irregularities;
  • other potential integrity concerns.

Automated indicators must be interpreted in context and, where necessary, followed by independent human assessment, appropriate clarification, or additional evidence.

Authors must not be accused of inappropriate AI use, fabrication, plagiarism, or other misconduct solely on the basis of an automated detection score or classification.

14. Prohibited Uses

AI and generative tools must not be used to:

  • fabricate or falsify data, results, references, ethical approvals, consent statements, reviewer reports, or other scholarly information;
  • disguise plagiarism or inappropriate reuse;
  • misrepresent generated or synthetic content as empirical evidence;
  • conceal material alterations of data or images;
  • fabricate reviewer identities or manipulate peer review;
  • generate false declarations concerning authorship, funding, competing interests, data availability, research ethics, informed consent, or clinical-trial registration;
  • compromise confidential manuscripts, peer-review materials, or protected research data;
  • impersonate another person in scholarly or editorial communication;
  • circumvent legitimate editorial, ethical, integrity, security, or confidentiality controls;
  • transfer responsibility for scholarly or editorial decisions from accountable human participants to automated systems.

Use of AI in connection with fraudulent, deceptive, manipulative, or otherwise integrity-compromising conduct may also be addressed under the Publication Ethics and Research Integrity policy.

15. Intellectual Property, Attribution, and Third-Party Rights

Users of AI tools remain responsible for ensuring that resulting content complies with applicable copyright, licensing, attribution, confidentiality, data-use, and intellectual-property requirements.

The use of an AI system does not guarantee that an output is original, lawfully reusable, appropriately attributed, or free from third-party rights.

Authors must verify that material included in a manuscript can lawfully and ethically be published and licensed under the terms applicable to the article.

AI tools must not be used to conceal or misrepresent the origin of copyrighted, licensed, confidential, proprietary, or otherwise protected material.

16. Bias, Limitations, and Scientific Interpretation

AI systems may produce incomplete, inaccurate, biased, inconsistent, unstable, or apparently plausible but unsupported outputs.

Researchers must critically evaluate such limitations where AI materially contributes to research methods, analyses, results, or interpretation.

Where relevant to the validity, reliability, reproducibility, fairness, or generalizability of the study, authors should report important limitations associated with:

  • training data;
  • model development;
  • model performance;
  • external or internal validation;
  • representativeness;
  • bias;
  • uncertainty;
  • robustness;
  • generalizability;
  • other characteristics materially affecting interpretation of the AI method.

AI-generated outputs must not substitute for appropriate scientific validation, methodological assessment, or human interpretation.

17. Concerns and Non-Compliance

Concerns about AI use must be assessed according to their nature, seriousness, transparency, available evidence, and effect on authorship, confidentiality, research methodology, reliability, or the integrity of the work.

Editors may request:

  • clarification of how an AI tool was used;
  • an AI-use disclosure statement or updated disclosure;
  • methodological information;
  • original or supporting data;
  • relevant source files;
  • code or analytical information;
  • confirmation and verification of references;
  • additional expert review;
  • other information necessary to evaluate the concern.

Failure to disclose material AI use where disclosure is required, provision of materially misleading information, fabricated content, confidentiality breaches, or other misuse may result in appropriate editorial action.

Not every undisclosed, inappropriate, or disputed use of an AI tool constitutes research misconduct.

Actions must be evidence-based, fair, and proportionate to the effect of the issue on authorship, transparency, confidentiality, scientific validity, research integrity, and the scholarly record.

18. Post-Publication Issues

Credible concerns concerning AI-assisted or AI-generated content identified after publication must be evaluated according to the same principles of fairness, evidence, proportionality, and protection of the scholarly record that apply to other publication-integrity concerns.

Where the underlying work remains reliable and the primary problem concerns incomplete disclosure, attribution, or reporting, clarification or correction may be sufficient.

Where AI-related practices materially undermine the authenticity, validity, provenance, reliability, ethical acceptability, or integrity of the published work, appropriate action may include:

  • correction;
  • expression of concern;
  • retraction;
  • institutional referral;
  • another proportionate measure necessary to protect the scholarly record.

Post-publication action must be taken in accordance with the Publication Ethics and Research Integrity and Corrections, Retractions and Expressions of Concern policies.

19. Relationship with Other Policies

This policy should be read together with applicable AMG Transcend Association and journal policies concerning:

  • Authorship and Contributor Responsibilities
  • Publication Ethics and Research Integrity
  • Data Integrity and Reproducibility
  • Image Manipulation
  • Peer Review Principles
  • Research Ethics
  • Human and Animal Research Ethics
  • Conflict of Interest and Competing Interests
  • Editorial Independence and Governance
  • Corrections, Retractions and Expressions of Concern
  • Complaints and Appeals

Individual journals may establish additional or more restrictive requirements concerning the use, disclosure, documentation, and assessment of AI and generative tools, provided that they remain consistent with the publisher-level principles established in this policy.