UNESCO expands AI data-governance guidance across Latin America
A Spanish-language UNESCO toolkit and new training resources aim to help governments manage data responsibly as artificial intelligence spreads.
Photo credit and license
Photo: Photo: Pilar / Wikimedia Commons, CC BY-SA 4.0. Archive 2 February 2020. · Image source · License: CC BY-SA 4.0UNESCO has expanded its guidance on responsible data governance in Latin America and the Caribbean with a Spanish-language edition of its Data Governance Toolkit and new training and self-assessment resources. The agency announced the development on 6 November 2026, one year after the toolkit began its international rollout.
The central issue is not simply building larger datasets. Governments must decide who can collect information, how it can be shared, what protections apply and who remains accountable when data is used to train or operate artificial-intelligence systems. UNESCO says clearer rules can help public institutions gain the benefits of digital services without neglecting privacy, inclusion and human rights.
Peru illustrates the policy dimension of the effort. The country published a National Artificial Intelligence Strategy for 2026–2030, drawing in part on UNESCO’s earlier readiness assessment. It also adopted a National Data Governance Strategy and action plan for the same period. These measures create frameworks for implementation; the announcement does not establish that all proposed safeguards are already operating effectively.
Training is another focus. UNESCO and UNDP said their course on inclusive digital and AI data governance received more than 1,000 applications in its first two cohorts in 2026. Government and regulatory personnel accounted for 84 percent of applicants, while the second cohort involved participants from 22 countries across the region.
The programme also includes an interactive institutional self-assessment tool developed with The GovLab, as well as work supporting community-led approaches to Indigenous and minority-language data. These components reflect a broader concern: systems trained on poorly governed data can exclude communities that are already underrepresented.
The expanded resources are practical tools rather than a binding regional law. Their value will depend on whether governments turn recommendations into transparent procedures, effective oversight and meaningful protections for the people whose information is used.