Declaração CRediT-IA and Ordinance 2664/2026 CNPq

In line with the new Policy on Integrity in Scientific Activity by CNPq, established by Ordinance No. 2,664/2026, Virtualia Journal presents the CRediT-IA Declaration model to the academic community. This unprecedented instrument, developed by the journal, is designed to transparently document the use of Generative Artificial Intelligence (GAI) tools in scientific manuscripts. The initiative was launched on the same day the ordinance was published—March 6, 2026—and is now available to authors who wish to align their submissions with the new scientific integrity guidelines.

CNPq Ordinance No. 2,664/2026, published in the Brazilian Federal Official Gazette on March 11, establishes that researchers must declare the use of Generative Artificial Intelligence tools of any kind and at any stage of research development, including conception, writing, data analysis, and submission, specifying in the respective texts and electronic presentations the tool used and its purpose. The document also prohibits submitting content generated by GAI as if it were human authorship, determining that authors are fully responsible for the final content, including any plagiarism or inaccuracies generated by the GAI.

The CRediT-IA model developed by Virtualia Journal meets these requirements by structuring a detailed declaration. The form requests manuscript identification—title, authors, and institutions—and asks whether AI tools were used. If so, the author must specify which tools were used, including their versions or models, and describe the purposes of use, such as technical support, language revision, text organization, or argumentative refinement. The declaration also requires that the limits of use be made explicit, clarifying that AI did not replace human authorship in fundamental intellectual decisions—such as defining the central thesis, selecting the theoretical framework, and making essential argumentative choices. Finally, the author signs a human responsibility statement, committing to having critically reviewed all content and assuming full responsibility for the final text.

The CRediT-IA Declaration adopts a format inspired by the CRediT (Contributor Roles Taxonomy) model, widely used for authorship definition, adapted to the reality of AI tools. By completing it, the author not only meets regulatory requirements but also contributes to a culture of methodological transparency and good scientific practices.

The creation of this model by Virtualia Journal represents a significant advance for the Brazilian academic community, offering a standardized instrument that facilitates compliance with the new CNPq requirements and strengthens research integrity in the age of artificial intelligence. The initiative aligns with the objectives of the CNPq Integrity Policy, which seeks to stimulate the quality and integrity of information at all stages of research projects, from conception to dissemination of results. The CRediT-IA model is now available for submission on the Virtualia Journal website, reaffirming the journal’s commitment to transparency and ethical responsibility in scientific production.