Brazilian Internet Committee Proposes 46 Guidelines for Social Media and AI Regulation

The Brazilian Internet Steering Committee, known as CGI.br, presented 46 guidelines on September 17, 2026, to guide the regulation of social media platforms in the country. The document addresses artificial intelligence, algorithms, competition, and platform responsibility. These recommendations are intended to serve as a reference for regulatory discussions and the formulation of applicable rules for digital platforms in Brazil. The guidelines do not create new legal obligations but aim to structure future regulatory frameworks.

The 46 guidelines are organized into six main axes: national sovereignty and jurisdiction, transparency, economy and competition, human rights, prevention and responsibility, and asymmetric governance and regulation. This publication expands upon ten principles for platform regulation that the committee presented in 2025. The earlier principles already advocated for transparency regarding boosting, distribution, moderation, and algorithmic recommendation mechanisms, as well as the opening of relevant data for researchers and authorities. The current recommendations result from debates conducted by the committee's Working Group on Platform Regulation and incorporate contributions from public consultations, including a 2023 consultation on digital platform regulation that received more than 1,000 submissions.

A primary focus of the document is the systems and algorithms used to determine content visibility for users. The guidelines suggest that platforms should make public the parameters employed in content recommendation algorithms and explain how personalization choices affect curation. Examples of data used by these algorithms include location, age, gender, and liked publications. The proposal does not necessarily require the disclosure of source code or information protected by trade secrets. Instead, the committee recommends that a potential regulatory authority define different levels of access, distinguishing between public information and data available only to regulators or researchers. During the launch event, a committee member highlighted the difficulty the academic community faces in analyzing social media operations due to current data access limitations. The proposal seeks to enable Brazilian researchers to investigate phenomena on platforms, with Brazilian entities participating in the certification of authorized researchers.

For professional users and businesses that rely on platforms for products and services, the recommendations extend to access information on ranking and display criteria, audience metrics, performance measurement tools, pricing, remuneration, and fees. The document also proposes that users be able to disable recommendations based on personal data profiling. Alternatives include chronological feeds or those organized by non-behavioral criteria, such as topics chosen by the user or accounts they have decided to follow.

Artificial intelligence is directly addressed in the recommendations. One guideline establishes that platforms must have the capacity to identify and label synthetic content, defined as material generated or significantly altered by AI. This identification should not rely exclusively on the declaration of the publisher. However, authors and users should have mechanisms to contest incorrect classifications. The committee also noted that algorithms and AI models used by platforms may be trained on large volumes of data that do not necessarily represent Brazilian cultural, social, demographic, educational, and linguistic realities. The document recommends audits to identify algorithmic biases and discrimination, along with periodic system testing. The goal is to consider risks during technology development, not only after problems arise.

The committee also advocates for human participation in content moderation, despite advances in automated systems. Moderation teams must master Portuguese and understand the Brazilian social, political, and cultural context. The composition and working conditions of these professionals must also be considered. The recommendations include mechanisms for identifying synthetic content, fact-checking tools, and measures for the protection of children and adolescents. Additionally, the guidelines address interoperability, a discussion with potential impact on platform architecture and business models.