The temporary committee that analyzes the bill that creates the regulatory framework for artificial intelligence (AI) in Brazil may vote on the text this Tuesday (3). If the report by senator Eduardo Gomes (PL-TO) is approved by the committee, the president of the Senate, Rodrigo Pacheco (PSD-MG), will be able to vote on the project in plenary on Thursday (5).

This vote was originally scheduled for December 2023, but pressure from big tech and disagreements regarding the text led to postponements. Last Thursday (28), the rapporteur presented a new version of the text which, according to him, strikes a balance between the preservation of fundamental rights and the guarantee of conditions for technological development.

“We are facing the so-called Collingridge dilemma here: regulating too early can stifle innovation, but regulating too late can allow irreparable damage,” said Eduardo Gomes. “AI has immense innovative potential capable of boosting our economic and social development. However, we also recognize its potential to cause harm on an equal scale.”

Among the changes made to the text, the so-called “exceptional hypotheses” were introduced to the legislation, aiming to exclude from its application certain uses carried out by individuals, non-profits, as well as activities aimed at testing and development.

The focus of governance measures ended up being on technologies considered “high risk”, starting to differentiate between startups and micro and small companies. Additionally, the protection of copyright, in addition to intellectual property and commercial and industrial secrecy, were included as foundations of the law.

“The evaluation criteria will not only be based on the scale of use, but will also consider the protection of fundamental rights, such as freedom of expression”, said the rapporteur. “This review allows for more detailed and attentive regulation, which understands the complexity of the social impacts of artificial intelligence. Integrity of information that does not take into account the need for freedom of expression with responsibility is inseparable.”

The changes caused entities representing productive sectors, such as the Federation of Industries of São Paulo (Fiesp) and the National Confederation of Industry (CNI), which previously opposed the project, to start defending the approval of the text.

The bill to regulate the use of artificial intelligence in Brazil was submitted in May 2023 by the president of the Senate, Rodrigo Pacheco, based on a draft prepared by a commission of jurists.

Pacheco maintains the expectation that the text will be approved in 2024.