Authorship

Authorship Criteria

Authorship criteria vary across frameworks, disciplines, and institutions, though most share the same basic structure. The International Committee of Medical Journal Editors (ICMJE) recommends four criteria [1], all of which must be met: making a meaningful contribution to the work's design, or to data collection, analysis, or interpretation; drafting the work or reviewing it critically for important intellectual content; approving the final version; and agreeing to be accountable for the whole work, ensuring that questions about its accuracy or integrity are properly investigated and resolved. ICMJE adds that anyone who meets the first criterion should also have the chance to take part in reviewing, drafting, and approving the final manuscript. Other frameworks follow the same logic with small variations. The US National Institutes of Health (NIH), for example, combines drafting and final approval into a single criterion, requiring three elements rather than four [2]. The Committee on Publication Ethics (COPE) recognizes this shared structure as the minimum standard across authorship definitions: substantial intellectual contribution; approval of the version; and accountability for the published work [3]. Where frameworks differ, it is usually in how specific or detailed they are; some publishers, for example, have expanded the first criterion to include creating new software used in the work [4]. Across all these frameworks, contributors who do not meet the full set of criteria are recognized in the acknowledgments section rather than listed as authors.

What Counts as a Contribution

These criteria were developed mainly for manuscripts and presentations, but the principles of credit and accountability apply to other research outputs too. What counts as a contribution worthy of authorship varies widely across disciplines, cultures, and even research groups within the same field [5]. The Contributor Roles Taxonomy (CRediT) addresses this by offering 14 standardized contribution roles, including Data Curation, Methodology, Visualization, Validation, and Software, making individual responsibilities clear and easy to record [6]. Because CRediT describes contributions without specifying which qualify for authorship, where that line falls still depends on the discipline and framework applied, with no universal agreement [5]. Since no single set of criteria can cover all modern research contexts, a principle-based framework built on transparency, credit, and accountability is recommended over fixed thresholds [7].

AI Tools and Authorship

Despite the lack of universal consensus, major research institutions and publishers agree on one boundary: AI tools cannot be listed as authors. Authorship requires accountability, consent to publication, and responsibility for the integrity of the work — obligations considered uniquely human. Use of AI tools must be disclosed in the methods or acknowledgments section [2, 4, 8, 9].