Brazil's AI framework bill still awaiting Chamber vote, now pushed past October elections

A briefing from the Brazilian law firm Chambarelli Advogados, updated in October 2026, explains that Brazil still has no general AI law. The leading proposal, Bill 2338/2023, passed the Federal Senate in December 2024 and remains under review by a special committee of the Chamber of Deputies. The firm says the rapporteur has indicated that the vote will come after the October 2026 elections. Any change by the Chamber would send the text back to the Senate. Until then, existing laws such as the LGPD data protection law plus consumer protection rules, already apply to AI systems.

The briefing recounts that the bill was introduced in the Senate in 2023, building on the work of a commission of jurists, and was approved with amendments. During 2026 the expected Chamber vote was postponed several times, and in August the rapporteur said it should take place at year's end, after the elections. Negotiations have involved the Senate rapporteur.

As to content, the Senate text uses a risk-based model that the firm likens in spirit to the EU AI Act. Uses of excessive risk, such as certain manipulative techniques and some biometric surveillance, would be prohibited with exceptions, while high-risk systems in sensitive settings would face impact assessments, human oversight, documentation and transparency duties. A national governance system coordinated by the data protection authority ANPD would work with sector regulators. People would be told when they interact with AI and could contest some decisions, and rules would address copyright and remuneration for training data, which remains an open issue.

For companies, the firm suggests inventorying and classifying AI systems, confirming the lawful basis for any personal data in training, adding human review, documenting data sources and testing, and reviewing vendor contracts.