Artificial intelligence reignites debates in the Brazilian House of Representatives

On February 24 and 25, 2026, two new bills addressing aspects of artificial intelligence were submitted to Brazil’s House of Representatives. The legal framework for AI is already being discussed in the context of Bill No. 2338/2023, which is currently being processed by the House of Representatives. The new bills focus on structural principles for AI development in Brazil and establish specific requirements for high-impact systems.

Bill No. 2338/2023 reinforces the trend towards a risk-based regulatory model, with significant implications for companies, developers, and users of AI systems. The topic intersects directly with core compliance areas, particularly the Brazilian General Data Protection Law (LGPD), intellectual property (including text and data mining and copyright issues), and emerging AI-related rules in the telecommunications sector.

From a personal data protection perspective, Bill No. 2338/2023 aligns with the LGPD’s principles of risk-based governance. It emphasizes the rights of individuals affected by AI systems and strengthens transparency and human review mechanisms for automated decisions that may materially affect data subjects. The bill also reinforces the institutional role of the Brazilian Data Protection Authority (ANPD) in this regulatory ecosystem.

Regarding intellectual property, debates around Bill No. 2338/2023 have highlighted text and data mining (TDM) as an essential component in the training, retraining, and validation of AI models – particularly generative AI. Key issues under discussion include:

• Transparency in training through dataset summaries rather than item-by-item disclosure;
• Opt-out mechanisms for data subjects;
• Remuneration issues; and
• Strengthening of data subjects’ rights (to be informed, to opt out, and to be protected against uses that violate or unfairly compete with their works).

In the telecommunications sector, AI initiatives must also consider National Telecommunications Agency (Anatel) regulations, including:

• The General Telecommunications Services Regulation (RGST) approved by ANATEL Resolution No. 777/2025, which incorporates principles applicable to AI deployment in the sector;
• The Cybersecurity Regulation for Telecommunications (RCiber), approved by ANATEL Resolution No. 740/2020, establishes guidelines such as security‑by‑design, privacy‑by‑design, and data protection; and
• Forthcoming amendments to Anatel’s oversight rules for AI-based systems are also under consideration. These amendments, as discussed in Public Consultation No. 31/2025, aim to balance the effective oversight of high-risk applications with reducing unnecessary regulatory burdens.

Bill No. 704/2026

Bill No. 704/2026 aims to establish foundational priorities and guiding principles for the development, deployment, and use of artificial intelligence across public and private sectors in Brazil.

One of the proposal’s key provisions is that the national AI policy must be centered on combating corruption and reducing social and regional inequalities (Article 3).

Under the bill, AI use must:

• Prioritize the prevention and detection of corruption, fraud, and the misappropriation of funds.
• Promote transparency, integrity, and traceability in automated decisions.
• Contribute to reducing social, economic, racial, regional, and digital inequalities.
• Foster inclusion, equitable access to opportunities, and responsible economic development.

The bill also identifies priority AI application areas – including contract and transaction analysis, credit scoring, public procurement, and financial flow monitoring – for detecting atypical patterns that may indicate fraud, overpricing, or favouritism. It prohibits systems that reinforce inequalities or produce exclusionary effects and mandates social impact assessments.

Bill No. 762/2026

Bill No. 762/2026 introduces a specific regulatory framework for high-impact or critical AI systems, requiring them to undergo an Artificial Intelligence Impact Assessment (AIA). Key elements include:

• Rules on classification, safety, privacy, governance, certification, oversight, and liability;
• Cumulative design, implementation, and operational requirements;
• Mandatory certification and business registration before operating in Brazil; and
• ANPD’s authority to issue certifications and define technical requirements.

Regarding data subject rights, the bill guarantees:

• Notice regarding the use of AI systems.
• Access to explainability, consistent with trade secrets and security constraints.
• Human review mechanisms for potentially harmful automated decisions.

The proposal also amends the Brazilian Civil Rights Framework for the Internet and the Consumer’s Defense Code to require platforms and providers to retain logs for auditing and investigations and to publish transparency reports related to high-risk AI use. In consumer relationships involving high-impact AI systems, providers must:

• Inform users about AI deployment;
• Provide channels for human review of consequential decisions;
• Assume strict liability for damage resulting from system failures.

The bill also promotes cooperation among the ANPD, the Ministry of Science, Technology, and Innovation, and sectoral agencies, and it establishes a permanent technical advisory committee.

These new proposals were expected, as the Legislative Branch’s engagement with AI-related matters is anticipated to expand significantly starting in 2026.

Demarest’s Telecommunications, Media and Technology, Data, Privacy and Cybersecurity, and Intellectual Property, Technology and Innovation teams remain available for any further clarification.