Generative AI Policies

Background and Purpose

Generative Artificial Intelligence (Generative AI) is becoming an increasingly important part of scientific research and academic communication. AI-based technologies can assist researchers in exploring research ideas, developing computational solutions, improving academic writing, processing information, and presenting scientific findings. At the same time, these technologies introduce challenges related to the reliability of research outputs, originality, transparency, confidentiality, and the attribution of intellectual contributions.

The International Transactions on Artificial Intelligence (ITALIC) acknowledges the potential of Generative AI to support scientific innovation while maintaining the principle that research integrity must remain a fundamental requirement of scholarly publication. The journal encourages researchers to explore appropriate AI technologies without compromising the quality, validity, or credibility of their scientific contributions.

This policy defines the responsibilities and limitations associated with the use of Generative AI in the preparation, evaluation, and publication of manuscripts submitted to ITALIC. It is intended to guide authors, reviewers, editors, and editorial staff in making responsible decisions about AI-assisted activities. The policy emphasizes that AI technologies must remain supporting instruments rather than substitutes for human expertise, independent reasoning, and ethical responsibility. Every individual involved in the publication process remains accountable for their own decisions and contributions.

Scope and Definitions

This policy covers AI technologies capable of generating, transforming, analyzing, or assisting with the production of academic content and research materials. These technologies include large language models, AI writing assistants, automated code-generation systems, image-generation applications, and other tools that provide AI-assisted functionality.

The policy applies to the complete editorial lifecycle of manuscripts submitted to ITALIC, from research preparation and initial submission through peer review, revision, editorial assessment, and publication. Its requirements apply to authors, reviewers, editors, and editorial personnel, although the specific responsibilities differ according to their roles.

Because ITALIC publishes research across multiple areas of artificial intelligence, the policy also addresses AI-assisted scientific activities, including model development, algorithm implementation, computational experimentation, data processing, prediction, classification, image analysis, and performance evaluation. Examples include research involving deep learning, machine learning, fuzzy systems, intelligent decision support, computer vision, natural language processing, robotics, and AI applications in environmental sustainability and green technology.

For the purposes of this policy, AI-assisted research refers to scientific work in which an AI system contributes to a research activity, analytical procedure, computational implementation, or manuscript-preparation task. AI-generated content refers to material produced wholly or partly through an AI system. Disclosure refers to providing sufficient information about relevant AI use to enable readers and editors to understand its role. Human accountability means that the responsible individuals retain ownership of the scientific and ethical obligations associated with their work.

Provisions for Authors

1. Acceptable Applications of Generative AI

ITALIC permits the use of Generative AI as a supporting resource when preparing manuscripts and conducting research, provided that the use is scientifically appropriate and does not undermine the authors' intellectual contribution. AI tools may help authors improve the readability of academic writing, correct language errors, translate text, organize preliminary ideas, and structure research documentation.

In technical research, AI-assisted applications may also support programming, code debugging, algorithm implementation, data exploration, feature engineering, and the development of computational workflows. Authors may employ AI-based methods directly within their research when these methods form part of the stated scientific objectives and are evaluated using appropriate procedures.

The acceptability of an AI-assisted activity depends on how the technology is used and how its outputs are incorporated into the research. Authors must examine generated material carefully, verify its accuracy, and revise it where necessary. AI-generated explanations, citations, code, mathematical expressions, and analytical suggestions must not be accepted without appropriate scrutiny.

When an AI system contributes to the actual research methodology, the authors must explain its function and relevance to the study. The description should be sufficient to distinguish the work performed by the researchers from the assistance provided by the technology. Authors must report the methods and results that were genuinely obtained rather than describing procedures suggested by an AI system as though they had been performed experimentally.

2. Unacceptable Applications of Generative AI

The use of Generative AI is unacceptable when it produces or facilitates misleading scientific claims, obscures the origin of research contributions, or compromises the integrity of scholarly communication.

Authors must not use AI systems to invent experimental observations, fabricate datasets, manipulate computational outcomes, or create statistical findings that do not originate from the stated research procedures. Generating fictitious scholarly references, incorrect DOIs, nonexistent publications, or unsupported quotations and presenting them as authentic sources is also prohibited.

AI technologies must not be used to disguise plagiarism, reproduce another researcher's work without appropriate acknowledgment, or present automatically generated scientific reasoning as verified original analysis. Authors must not manipulate model configurations, evaluation metrics, benchmark results, or reported performance to create an inaccurate impression of a proposed method's effectiveness.

Particular care is required when AI is used to generate or modify figures, images, graphs, and other visual materials. Such materials must not distort the underlying evidence or falsely represent experimental conditions, observations, or model performance. AI-assisted content must also comply with applicable copyright, privacy, intellectual property, and research ethics requirements.

Authors remain responsible for ensuring that their manuscripts represent genuine scholarly contributions. The use of an AI system does not justify inaccurate reporting, inadequate verification, or the misrepresentation of research activities.

3. Transparency and Disclosure

Transparency is essential whenever Generative AI makes a substantive contribution to research communication or manuscript preparation. Authors must disclose relevant AI use when a tool generates or substantially revises manuscript content, contributes to the interpretation of research information, supports the production of research-related visual materials, or performs other activities that materially affect the work.

The disclosure should identify the tool used, explain the purpose of its application, and indicate the relevant stage of the research or writing process. Authors should also explain how AI-assisted outputs were reviewed and verified. Where an AI system forms part of the research methodology, its role must be described in the Methods or Methodology section, together with other information necessary to understand the research procedure.

For substantive assistance with manuscript preparation, ITALIC recommends including a declaration before the References section under the heading "Declaration of Generative AI and AI-Assisted Technologies", unless the journal's submission instructions specify another location.

Authors may use the following statement as a guide, adapting it to accurately describe their circumstances:

"During the preparation of this manuscript, the authors used [AI tool and version, where available] to assist with [specific activity]. The resulting material was critically reviewed and verified by the authors, who remain fully responsible for the content, accuracy, and integrity of the submitted manuscript."

Routine spelling correction and basic grammar assistance that does not generate substantive content generally do not require a separate declaration. Authors must not provide a disclosure that exaggerates, minimizes, or misrepresents the actual role of AI in their work.

4. Human Responsibility for Research Outputs

The use of Generative AI does not reduce the responsibilities of authors for the research they submit to ITALIC. Authors must ensure that their work is scientifically defensible, accurately documented, and consistent with accepted research and publication ethics.

Before submission, authors must verify the accuracy of factual statements, the authenticity of references, the validity of analytical procedures, and the consistency of reported results with the underlying evidence. For computational studies, this responsibility includes checking code behavior, model configurations, data-processing procedures, evaluation metrics, and the interpretation of experimental outcomes.

Authors must also consider whether AI-assisted procedures introduce bias, unsupported assumptions, security vulnerabilities, or limitations that could affect the findings. Relevant limitations must be acknowledged when they influence the interpretation or reproducibility of the research.

AI systems cannot be listed as authors or co-authors because they cannot independently accept the obligations of scholarly authorship. Human authors must make the intellectual contributions required by the journal's authorship criteria, approve the final manuscript, and accept responsibility for the published work.

Provisions for Reviewers

1. Protection of Unpublished Research

Peer review requires reviewers to handle unpublished research with discretion and professional responsibility. Manuscripts submitted to ITALIC may contain original algorithms, experimental results, datasets, source code, technical designs, or other intellectual contributions that must not be disclosed without authorization.

Reviewers must not transfer confidential manuscript materials to public AI platforms or external AI services that have not been authorized by the journal and assessed for appropriate confidentiality and data protection. This restriction includes uploading manuscript files, copying substantial excerpts, sharing supplementary data, or submitting confidential review materials for automated summarization or evaluation.

Reviewers must also avoid using AI services in ways that could expose unpublished research, compromise author privacy, or allow confidential information to be retained or reused without authorization. The confidentiality obligation remains in effect throughout and after the review process.

2. Appropriate AI Assistance

Reviewers may use limited AI assistance for general language improvement or non-confidential administrative tasks, provided that such activities do not involve unauthorized disclosure of manuscript content or interfere with the independent assessment of the research.

Any permitted AI-assisted material must be checked before it is included in a review report or editorial communication. Reviewers must ensure that the resulting comments accurately reflect their own assessment and do not introduce unsupported claims, technical inaccuracies, or misleading statements.

ITALIC expects reviewers to exercise particular caution when evaluating studies involving complex computational models, novel algorithms, statistical procedures, or claims of improved predictive performance. An AI-generated explanation cannot establish that a method is valid, that an experiment is reproducible, or that a reported result is scientifically justified.

3. Independent Evaluation

Reviewers are responsible for forming their own judgments about the scientific quality and suitability of each manuscript. Generative AI must not replace the reviewer's expertise in assessing originality, methodological soundness, technical contribution, experimental design, validity of results, and relevance to the journal's scope.

Reviewers must not delegate manuscript evaluation or the formulation of substantive recommendations to AI systems. Acceptance, revision, and rejection recommendations must reflect an independent evaluation of the manuscript and its supporting evidence.

If reviewers identify possible research misconduct, unreliable results, inappropriate data manipulation, or other ethical concerns, they must communicate the issue to the editor using evidence from their own assessment. Automated outputs or AI-generated suspicions must not be treated as sufficient proof of misconduct.

Provisions for Editors

1. Confidentiality in Editorial Management

Editors and editorial staff are responsible for protecting the confidentiality of manuscripts, reviewer identities where applicable, review reports, author information, and editorial correspondence. The use of AI-assisted services must not expose these materials to unauthorized parties.

Confidential manuscript content must not be submitted to public AI platforms or external AI services unless the journal has explicitly authorized the service and established appropriate safeguards. This requirement applies to editorial screening, manuscript summarization, reviewer-report processing, and other activities involving unpublished research.

Editors must consider the confidentiality, privacy, and data-handling practices of any AI service before authorizing its use in editorial operations. The convenience of automated processing must not take precedence over the protection of authors' intellectual contributions or the integrity of the editorial process.

2. AI-Assisted Editorial Activities

AI tools may support selected administrative activities, such as improving the language of general journal communications, organizing non-confidential workflow information, and maintaining clear descriptions of editorial procedures. These applications may help improve efficiency but must remain subject to human review.

Editors and editorial staff must verify AI-assisted communications before sending them to authors, reviewers, or other parties. They must ensure that automated assistance does not introduce inaccurate instructions, inappropriate statements, or inconsistencies with the journal's policies.

Any use of AI in editorial administration must be appropriate to the task and must preserve impartiality, confidentiality, and professional accountability. AI tools must not be used to bypass established editorial responsibilities or substitute automated judgments for necessary human decisions.

3. Editorial Independence and Decision-Making

Editorial decisions must be made by responsible human editors based on the manuscript's scientific contribution, methodological quality, relevance to ITALIC, reviewers' assessments, and applicable publication ethics requirements.

AI-generated scores, summaries, recommendations, or classifications must not serve as the sole basis for determining whether a manuscript should be accepted, revised, or rejected. Editors must independently examine the relevant evidence and apply their professional judgment to the circumstances of each submission.

Editors must also ensure that authors are treated fairly and that editorial decisions are not based on unsupported assumptions about the use of AI. Where concerns arise, the editor should request clarification or supporting evidence before determining the appropriate course of action.

4. Assessment of Suspected AI Misuse

AI-detection technologies may assist editors in identifying material that warrants further examination. However, such technologies have limitations and may produce inaccurate classifications. Their outputs must therefore be interpreted cautiously and cannot independently establish that an author has violated this policy.

When a potential violation is identified, the editor must consider the available evidence, the relevant disclosure requirements, the nature of the suspected activity, and the effect on research integrity. Authors should be given an appropriate opportunity to explain their use of AI when clarification is necessary.

Any resulting editorial action must be proportionate to the evidence and consistent with the journal's publication ethics procedures. The objective is to protect the reliability of the scholarly record while ensuring a fair and transparent assessment of each case.

Violations and Consequences

ITALIC considers compliance with this policy an essential part of responsible scholarly publication. Potential violations include misleading disclosures, fabrication of AI-assisted research outputs, submission of fictitious references, inappropriate manipulation of scientific evidence, unauthorized disclosure of confidential manuscripts, and other uses of AI that compromise research or publication integrity.

When a concern arises, the journal may request an explanation from the relevant parties and examine supporting information, including original research files, datasets, code, model configurations, processing records, or other documentation appropriate to the case.

The response will depend on the seriousness of the conduct, the available evidence, whether the violation affects the validity of the research, and whether the issue can be resolved through clarification or correction. Possible measures include requesting a revised disclosure, requiring manuscript changes, rejecting a submission, publishing a correction, or retracting an article when necessary to protect the scholarly record.

ITALIC will seek to handle suspected violations consistently and fairly. Any action must be based on an assessment of the circumstances rather than an unsupported assumption or an automated AI-detection result alone. Matters involving serious concerns about research integrity will be handled in accordance with the journal's applicable publication ethics procedures.

Policy Review

The development of Generative AI continues to change the ways research is conducted, evaluated, and communicated. ITALIC will periodically assess this policy to ensure that its requirements remain appropriate for emerging technologies, evolving research practices, and developments in scholarly publishing ethics.

Revisions may be introduced when necessary to clarify disclosure requirements, address new risks to research integrity, improve the protection of confidential information, or respond to updated guidance from recognized scholarly publishing organizations.

The version published on the official ITALIC website will serve as the journal's applicable policy. Authors, reviewers, editors, and editorial staff are responsible for consulting and following the current requirements relevant to their roles.

References and Guidelines

The development of this policy is informed by established principles of research integrity and guidance on the responsible use of artificial intelligence in scholarly communication. The following resources provide additional information for individuals involved in research and academic publishing:

These resources provide general guidance and do not replace the specific requirements established by ITALIC. Where necessary, the journal will interpret and update its policy in light of developments in AI research, publication ethics, and applicable scholarly publishing standards.