Bairo GonzalezLeandro · Martinez
Start your guided journey The story

In the world · Artificial intelligence · 24 August 2026

Bill 2338/2023: where Brazil's AI framework stands in 2026

Passed by the Senate in 2024, Bill 2338/2023 is still in the Chamber and the vote was pushed past the elections. What it provides and what changes.

bairogonzalez.com team, drawing on Bairo's story · Published

Bill 2338/2023 (Projeto de Lei 2338/2023), Brazil's legal framework for artificial intelligence, was passed by the Senate on December 10, 2024 and, as of October 2026, is still in the Chamber of Deputies. On August 24, 2026, the rapporteur, Congressman Aguinaldo Ribeiro (PP-PB), said the vote will come after the October elections. Once the Chamber is done, the text goes back to the Senate.

What happened

The bill originated in the Senate, authored by Senator Rodrigo Pacheco, building on the work of a commission of legal experts. The substitute text approved in December 2024, by voice vote, organizes AI by degree of risk, according to Agência Senado:

  • Prohibited uses, such as autonomous weapons, subliminal techniques that exploit vulnerabilities and systems that try to predict crime from a person's personality.
  • High risk, with more duties, for uses such as hiring, access to education, autonomous vehicles and criminal investigation tools.
  • A national system of regulation and governance, coordinated by the ANPD (Brazil's data protection authority), together with each sector's regulators.
  • Copyright: commercial use of protected works to train AI will require paying the rights holders, with limited exceptions for research.

In the Chamber, the text got a special committee, chaired by Congresswoman Luisa Canziani (PSD-PR), with Aguinaldo Ribeiro as rapporteur, according to Agência Câmara. The vote was at one point expected in the first half of 2026 and did not happen. At Febraban Tech 2026, the rapporteur told Mobile Time that the work will resume "at full force" after the elections, and that copyright and the other open issues are being handled jointly with the Senate.

Why it matters

Until the law comes out, Brazil uses AI under the rules that already exist: the LGPD (Brazil's General Data Protection Law), the Consumer Protection Code, the standards of each professional council and the TSE (Superior Electoral Court) resolutions for the election. For the small business owner, that means two things. First, there is no general compliance "stamp" to ask for today. Second, the direction is already clear in the Senate text: the greater the risk to someone's life, job and rights, the greater the duty to explain, keep records and leave a human in charge.

The delay also has a useful side. It gives those who build AI time to design the product with these rules in mind from the start, instead of patching it later.

In Bairo's view

For Bairo Leandro Gonzalez Martinez, the rule the bill pursues is already the right rule: the machine organizes, the person decides. It is the same idea he repeats when he says artificial intelligence is an extension of the human, not the problem. A law that requires a responsible human in high-risk uses, in his reading, does not slow AI down; it protects what AI is supposed to serve.

He also sees in the framework an argument in favor of niche AIs. An AI that does everything is hard to classify by risk. An AI that does one thing, for one audience, with a human at the end of the line, is easy to explain to a regulator. That is why CLAIN, the agent orchestrator he founded, is designed as a team of small specialists rather than a single mind. The goal of 142 niche AIs in 142 days, which he announced on October 1, 2026, is a goal, and each of them will have to fit this risk yardstick.

Where this meets the ecosystem

The ventures Bairo founded touch precisely the areas the bill treats as sensitive, and the proposal is to handle them with the care the text asks for:

  • In law, SOSPROCESSOS is an analysis tool for lawyers, under development: it organizes rulings and scenarios, and the lawyer is the one who signs the defense.
  • In health, Hospital Triage proposes pre-consultation with a clinical assistant and validation by a professional, which is the design the framework and the CFM (Federal Council of Medicine) call for.
  • In agribusiness, CEASA Bank, a financial platform for agribusiness that operates on the infrastructure of an authorized institution, proposes using AI to quote cargo prices, a lower-risk use than a credit decision, which calls for more explanation.

None of this depends on the law coming out tomorrow. It depends on building now as if it were already in force.

Sources

Read also

More on this topic

In the world: Artificial intelligence