If approved by the Senate, bill will still require review by the Lower House before it can be sent for presidential approval.
The latest proposal for regulating artificial intelligence (AI) in Brazil could be approved next Tuesday (3) by the Senate’s temporary committee, which was established to address the matter. The proposal could then be brought to a vote in the Senate plenary later that week.
The timetable for the committee’s vote was announced on Thursday (28) by the committee’s chair, Senator Carlos Viana. The possibility of the proposal passing in the Senate was indicated by the legal framework’s rapporteur, Senator Eduardo Gomes, following the reading of the latest report.
During a committee meeting, Mr. Viana explained that requests for a review of the proposal, which could delay the vote again, are no longer allowed. According to him, this phase is meant only for “report supplements,” and the submission of new amendments is still permitted.
If approved in the Senate, Bill 2338/23 will still need to be reviewed by the Lower House before proceeding to be signed into law.
The rapporteur’s new report retains the structure of previous reports. It includes, for instance, a proposal to create a National System for AI Regulation and Governance (SIA) to oversee AI use in the country. Under this system, the National Data Protection Authority (ANPD) would coordinate the actions of regulatory agencies in each sector.
Throughout the debate, lawmakers and experts were divided on whether to merely establish principles for AI use or create more specific regulations for different issues. The challenge lies in how to mitigate the risks of AI systems without stifling innovation and investment in the sector.
Attorney Patricia Peck, CEO of Peck Advogados, believes the proposal still needs improvements. After reviewing the report, she noted that it contains aspects that are “difficult to implement in the national context.” The digital law expert highlights that the text does not, for example, establish mechanisms to combat unauthorized use of image and voice in AI image manipulation, within the concept of deepfake.
Mr. Gomes argued that, within its governance framework, the current text protects fundamental rights, ensuring that “automated decisions are not made in isolation, but with appropriate oversight.”
Source: Valor International
