The preliminary report on AI regulations presented to Brazil’s Senate last week provides a middle-of-the-road solution between two previous bills on the topic. It is also more flexible and innovative than the European Union’s AI Act, from which it draws inspiration, providing rights for people affected by AI systems — something unusual whether in more restrictive legislation such as the European law or a more principled discussion like in the U.S.
The bill is only the first step of the process, providing a concrete basis for more debate. Senators can suggest amendments to the proposal until May 9. Several entities and experts are also mobilizing to influence legislators in the coming weeks, as Senate President Rodrigo Pacheco — author of one of the previous AI bills — chose the topic as a priority for 2024.
Despite being one of the first countries to start discussing AI regulations, even before the European Union, in 2020, Brazil’s debate on the issue has changed direction, from a regulation proposal deemed too broad (Bill 21/2020) presented that year, following the country’s EBIA artificial intelligence strategy, to a replacement proposal (Bill 2338/2023) drafted by a group of legal experts aimed at introducing a full-fledged regulatory framework.
“The text presented last week lies somewhere between these two proposals, but closer to the latter, I would say, as it maintains a risk-based regulatory architecture, just like the EU proposed, but in a more flexible way,” says Marcela Mattiuzzo, a data protection lawyer and also a columnist at The Brazilian Report. The text results from a review of Bill 2338 by a temporary committee set up by Mr. Pacheco.
The intention of the Senate president was for the matter to be approved by the end of 2023 so that it could be applied in this year’s municipal elections, but that...
