Brazil: AI Bill rapporteur proposes a smaller role for the ANPD

Brazil: AI Bill rapporteur proposes a smaller role for the ANPD

In brief

  • On 9 November 2026, Deputy Aguinaldo Ribeiro, rapporteur of Bill 2,338/2023 in Brazil's Chamber of Deputies, proposed reducing the ANPD's role in AI regulation.
  • His model, inspired by the National Monetary Council, would separate the setting of AI policy from the execution of regulation.
  • The bill has not yet been voted: the rapporteur wants a vote before year end, with parliamentary recess starting on 23 December.

The rapporteur of Brazil’s artificial intelligence bill proposes that the Autoridade Nacional de Proteção de Dados (ANPD) stop being the central body for rule-making and focus on enforcement. The debate goes to the institutional design of the future law, and so to the legal certainty of anyone developing or using AI in Brazil.

What the rapporteur has proposed

As reported by JOTA, Deputy Aguinaldo Ribeiro (PP-PB) says the agency would keep only the execution of the rules. According to published coverage of his remarks on Monday 9 November 2026, he regards deciding which body regulates as the main obstacle for Projeto de Lei 2.338/2023 in the Câmara dos Deputados.

He suggests a structure that splits the definition of AI policy from the execution of regulation, modelled on the Conselho Monetário Nacional (CMN). His argument is that this would yield regulation able to withstand technological change. He aims to vote the text before the end of the year, although parliamentary recess begins on 23 December.

Where we come from

The Senate approved Bill 2,338/2023 on 10 December 2024 and sent it to the Chamber in March 2025. The text provided for a Sistema Nacional de Regulação e Governança de IA (SIA) coordinated by the ANPD. In December 2025 the Executive submitted its own bill, which was attached to the main one. According to Convergência Digital, that text created a Conselho Brasileiro para Inteligência Artificial (CBIA) and gave the ANPD the role of residual regulator, with rule-making, supervisory and sanctioning powers in sectors without specific AI regulation. The change now floated would alter that balance.

What it means in practice

Until a general law exists, the ANPD continues to apply the LGPD to personal data processing by AI systems. A future law separating who sets the rules from who enforces them would require companies to work out more carefully which body they answer to for rules, inspections and sanctions. If the split is unclear, the risk of fragmentation between the ANPD, sectoral regulators and the new policy body grows.

Our view at CEA Digital Law

The Euro-American Committee of Digital Law - CEA Digital Law considers the debate on who should write AI rules a legitimate one, and the separation of policy and execution has precedents in financial regulation. For it to work, we suggest three safeguards.

  • Preserve the technical capacity the ANPD has built in data and algorithms, for example through binding technical opinions or formal cooperation with the rule-making body.
  • Define competences precisely, to avoid overlaps between data, sectoral and policy authorities.
  • Ensure participation by civil society, academia and affected people, as the planned advisory committees envisage.

Comparison is useful. Regulation (EU) 2024/1689 combines an AI Office, national authorities and coordination with data protection authorities, and several Latin American and African countries are moving towards mixed models. Brazil’s experience may show that the choice of institutional design matters as much as the content of the obligations. It remains open whether the final structure will keep a strong link with data protection and fundamental rights.

Sources

CEA Digital Law

Lettre d’information

Événements, actualités et analyses sur le droit numérique, dans votre boîte de réception. Seulement quand il y a quelque chose à dire.