From disclosure to traceability: editorial governance of artificial intelligence in Open Science

Abstract

Review of the book Inteligência artificial na editoração científica em Ciência Aberta: riscos, integridade científica e governança editorial, by Bernardo Dionízio Vechi, Milton Shintaku, Maria Aniolly Queiroz Maia and Rosilene Paiva Marinho de Sousa (Editora Ibict, 2026). The book maps the entry points of artificial intelligence into the editorial workflow and translates integrity principles into verifiable routines, organized by stage (submission, peer review, copyediting, metadata curation and post-publication). The review describes the structure of the book and assesses it from the perspective of information reliability, the semantic web and the use of AI agents. It highlights the central role given to metadata and traceability, the treatment of automated outputs as signals rather than verdicts, and the attention paid to hidden prompt injection in manuscripts. As limitations and agenda, it notes the lack of empirical validation of the model, the limited exploration of machine-readable representations of AI-use disclosures, the tensions between hidden probes and accessibility, and the role of journals as producers of the corpora that feed generative systems. It concludes that the book is a timely reference of high practical value for Ibero-American editors.