Brazilian Government Establishes Guidelines on Best Practices and Transparency in the Use of Artificial Intelligence in Education and Research

Brazilian Government Establishes Guidelines on Best Practices and Transparency in the Use of Artificial Intelligence in Education and Research

Recently, two relevant initiatives concerning the use of artificial intelligence (AI) in education and in the academic sphere were published in Brazil. In February 2026, the Ministry of Education (MEC) released the Framework for the Responsible Development and Use of Artificial Intelligence in Education, which establishes parameters for the use of AI from early childhood education through postgraduate studies. In March 2026, the National Council for Scientific and Technological Development (CNPq), through Ordinance No. 2,664/2026, instituted the Policy on Integrity in Scientific Activity, including specific guidelines on the use of Generative Artificial Intelligence (GAI) tools in research. Taken together, these initiatives reinforce the centrality of transparency and the responsible use of technology.

The MEC Framework, although non-binding in nature, is likely to guide the conduct of education systems, institutions, and providers of educational technology. One of the most relevant aspects of the document lies in the manner in which it articulates legal risks and governance mechanisms in the use of AI in concrete educational contexts. In the field of data protection, it proceeds from the premise that such technologies entail the intensive processing of data relating to students and teachers, including sensitive data, thereby requiring compliance with the General Data Protection Law, with particular emphasis on transparency, security, and the identification of appropriate legal bases.

The document also highlights the risks of algorithmic bias, particularly in applications such as performance assessment and the personalization of education, which may impact educational trajectories, as well as limitations in the reliability of generative AI systems. In this context, it establishes as a guiding principle that AI should function as a support tool, without replacing the pedagogical role of the teacher. In the field of intellectual property, the Framework acknowledges uncertainties regarding the use of protected content in the training of such systems and the potential reproduction of works in generated outputs, thereby signaling the need for caution on the part of institutions.

To address these risks, the document recommends the adoption of algorithmic impact assessments in applications that may affect the rights of students and teachers, encourages the strengthening of institutional capacity to understand and supervise the tools adopted, and refers to the use of experimental regulatory environments (sandboxes) for controlled testing.

Within the sphere of scientific research, Ordinance No. 2,664/2026 introduces direct obligations with immediate impact for researchers and institutions affiliated with the CNPq. The regulation establishes, in Article 9, that the use of generative artificial intelligence tools must be disclosed at any stage of the research, including conception, data analysis, and drafting, with an indication of the tool used and its purpose. Furthermore, the submission of content generated by GAI as if it were of human authorship is expressly prohibited, with authors remaining fully responsible for the final content, including any instances of plagiarism or inaccuracies generated by the tool employed. The Ordinance also prohibits the submission of third-party research projects into GAI tools for the preparation of scientific opinions.

Taken together, these initiatives reflect a broader movement in Brazil, emerging through the development of best practice guidelines for the use of AI tailored to specific sectors. In the absence of a comprehensive legal framework on artificial intelligence, sector-specific guidance is being developed and is already producing practical effects. By aligning themes such as transparency, human oversight, and risk assessment with international discussions, the country is beginning to outline its own model of AI governance. For educational institutions, researchers, and companies involved in developing or providing technological solutions for the education sector, this scenario indicates that the adoption of AI now requires not only innovation, but also the structuring of control mechanisms, documentation, and compliance.

The full text of the Framework for the Responsible Development and Use of Artificial Intelligence in Education is available at: https://www.gov.br/mec/pt-br/media/segape/referencial-oficial-pt.pdf

Ordinance CNPq No. 2,664/2026 can be accessed at: http://www.cnpq.br/web/guest/view/-/journal_content/56_INSTANCE_0oED/10157/23142775.

Note: For quick release, this English version is provided by automated translation without human review.