Ethics & Publication Malpractice Policy

This policy applies to all participants in the publication process of AI DECISIONS: authors, reviewers, editors, the publisher, and readers. We publish work on human–AI decision-making systems, with primary emphasis on decision quality, safety, fairness, and reproducibility.

Editors

• Ensure fair, unbiased, and timely consideration of manuscripts
• Make decisions based on scientific merit and journal fit
• Manage conflicts of interest and maintain confidentiality
• Follow COPE flowcharts for investigations and corrective actions

Authors

• Guarantee originality, data accuracy, and honest reporting
• Provide materials for reproducibility or a justified exception
• Disclose conflicts of interest and funding sources
• Comply with ethical and legal requirements

Reviewers

• Provide objective, confidential, and timely reviews
• Declare conflicts of interest and decline when conflicts exist
• Do not use unpublished materials for personal research

Publisher

Supports editorial independence and provides legal/procedural support for investigations and corrections.

• The journal uses the CRediT taxonomy. Authorship is based on substantial contributions and collective responsibility for the content.
• Authorship changes after submission require written consent of all co-authors (COPE guidance applies).
• AI tools are not authors. Their use must be disclosed (see section 7).

• Manuscripts must be original and not under consideration elsewhere.
• Preprints are welcome (provide the DOI/URL at submission).
• Textual overlap may be screened with appropriate tools.
• Self-citation should be moderate and justified.

A Data/Code Availability section is required. Persistent repositories with DOIs are preferred. Where possible, provide datasets, code and configurations, data/model cards, and protocols. Preregistration and Registered Reports are supported.

For work involving people or their data, IRB/IEC approval or a waiver is required, alongside compliance with applicable law (including UK GDPR/GDPR). De-identification and appropriate rights/consents are required for publication of images/medical materials.

Allowed:

Assistance with language/editing, code, and visualisation — provided human verification and disclosure in the manuscript.

Prohibited:

Generating/fabricating data, substituting experiments/results, creating research images without explicit labelling as synthetic, sending confidential manuscripts to external AI services without a legal basis. AI is not listed as an author.

All articles must provide a Model Card and Data Card (or combined System Card) that disclose purpose, boundaries, data, metrics, monitoring plan, and safety measures. For high-stakes domains, a High-stakes Checklist and red-teaming report are mandatory.

All authors, editors, and reviewers must disclose financial and non-financial conflicts. Funding sources and their role must be stated. Editors do not handle manuscripts where they have a conflict.

Default: double-blind peer review. The editorial office may conduct initial screening for scope, ethics, and quality. Reviewers must keep materials confidential.

• Errors affecting interpretation: Corrections (Erratum/Corrigendum) issued.
• Serious uncertainty: Expression of Concern may be published.
• Proven misconduct: Retraction with metadata retained and reasons stated.

Suspected fabrication, falsification, plagiarism, or duplicate submission is handled per COPE guidance. Possible actions: correction, rejection, retraction, notification of institutions, temporary submission ban.

Appeals against editorial decisions must be lodged within 14 days of notification and should cite factual/procedural errors. Complaints about ethics/conflicts should be sent to journal@aidecisions.ai. Receipt is acknowledged within 7 days.

Articles are published under CC BY 4.0. Authors retain copyright and grant the journal a non-exclusive licence. Third-party materials require permissions and attribution.

Personal data of authors/reviewers are processed in accordance with UK GDPR/GDPR. Manuscripts/reviews must not be uploaded to third-party services without a legal basis.

Advertising and sponsored content are clearly labelled and do not influence editorial decisions. Special issues follow the same standards.

All articles receive a DOI. Comments/replies are permitted after publication; substantive concerns trigger editorial assessment.

This policy is reviewed at least annually or when community norms change.