Brazil Moves to Let Agencies Define AI Risk as Congress Races to Vote in June

Overview

• Relator Aguinaldo Ribeiro has proposed removing fixed risk categories from the bill and leaving classification of high- and low-risk AI systems to regulatory agencies so rules can change with technology.
• Chamber president Hugo Motta said the relator’s committee report will be presented on June 9 and the Chamber aims to vote the AI bill in June before sending it back to the Senate for final approval.
• The version approved by the Senate in 2024 still underpins the process with stricter obligations for high-risk uses such as autonomous vehicles and medical diagnostics and an explicit ban on AI in autonomous weapons.
• Speakers at the Brasília Tech Summit, including PGR Paulo Gonet and CEOs from Amazon and Google, urged measured rules that protect rights while avoiding regulatory 'excesses' that could deter multi-year private investment.
• If the Chamber follows the June timetable the next steps are the special commission vote, a plenary vote in the Chamber, and then a return to the Senate, a sequence that will determine how adaptable versus prescriptive Brazil’s final AI framework will be.