Brazil Puts Its AI Ethical Impact Self-Assessment Online for More Than 250 Federal Bodies
Federal public bodies in Brazil now have an online tool to assess ethical risks, spot governance gaps and improve how they develop and use artificial intelligence systems, Capital Digital reported on November 13. The platform, called the Self-Assessment of Ethical Impact in Artificial Intelligence (AIE), was developed by the Ministry of Management and Innovation in Public Services (MGI) and is available to the more than 250 bodies that belong to the federal IT resources administration system, known as SISP.
The tool already existed but only worked offline. With the new web version, institutions can carry out the assessment over the internet and review their own practices to identify possible problems linked to their use of AI. The system looks at transparency, human oversight, fairness, security, privacy and accountability. Based on the answers, a body can identify good practices, gaps and risks, and receive recommendations for prevention and improvement measures. The intention is that the exercise does not stop at a diagnosis: the results are meant to guide changes in governance and to accompany different stages of developing and using AI systems in the public sector.
What it means in practice
The initiative is part of the Brazilian Artificial Intelligence Plan (PBIA) and received consultancy from the World Bank. The platform has been released as open source, so other institutions, including in other countries, can reuse and adapt it. The MGI’s Digital Government Secretariat is also preparing practical guides to help public bodies turn responsible AI principles into concrete measures. Alongside the tool, the government published a new page bringing together the initiatives of its Artificial Intelligence Unit (NIA), with information on training, indicators and ethical use, and projects such as Chat GOV.BR and the Brazilian Artificial Intelligence Catalogue (CBIA). Institutions involved include the Casa Civil, the Ministry of Science, Technology and Innovation, Finep, CPQD, Serpro, Dataprev, the National School of Public Administration (Enap), the World Bank and NIC.br.
The self-assessment adds a governance layer as Brasília pushes AI deeper into public administration, from the MGI’s plan for transactional AI in Contrata+Brasil to the Senate and Receita Federal training civil servants in AI. States are moving too, as Rio de Janeiro’s statewide AI policy with an ethics committee shows. What to watch is how many federal bodies actually complete the assessment, whether results are published, and whether it becomes mandatory before new AI systems go live.