Brazil's AI Law Was Designed to Create Certainty. It May Create the Opposite.

Brazil has 1,400 AI-first startups actively mapped across 424 cities — growth of over 40% in two years. Less than 3% of the capital funding them comes from abroad. For context, startups with international VC exposure grow four to ten times faster than their domestically funded peers. That gap is the number Brazilian AI founders should be most concerned about — and the AI regulatory debate now under way in Brasília will either close it or deepen it.

PL 2338/2023 — Brazil's AI regulatory framework — was approved unanimously by the Senate in December 2024. It returns to the Câmara dos Deputados in 2026 alongside a companion Executive bill that would create the Sistema Nacional de Regulação e Governança de IA (SIA). The stated goal is legal clarity. The structural risk is the opposite: a compliance architecture that disproportionately burdens the companies least capable of absorbing it.

The European parallel is instructive. VC investment in EU AI startups fell approximately 15% in 2024, with regulatory uncertainty cited as a contributing factor. The EU AI Act created compliance requirements that large technology companies could absorb and small companies could not. The firms most capable of navigating the compliance burden were the incumbents the regulation was supposedly designed to govern. That inversion — where regulation protects incumbents under the cover of protecting users — is precisely the risk that specialists in Brazil are flagging now.

The investment calculus for international capital is direct: if Brazil's regulatory environment resembles the EU's, but without the EU's established legal infrastructure, deep talent pool, and proximity to major markets, the capital goes elsewhere. International VC backing Brazilian AI companies needs predictability, not just the existence of rules. A framework that creates ambiguity about which applications qualify as high-risk — triggering conformity assessments that early-stage companies cannot afford — produces exactly the unpredictability that sends capital to more familiar jurisdictions.

The counter-argument deserves serious weight. Fintechs, credit underwriters, and health AI companies building in regulated industries face constant legal uncertainty about where their models sit within existing financial or medical regulation. A clear AI framework could be the permissive structure that lets regulated-industry AI scale in Brazil in ways it cannot elsewhere. The Banco Central has consistently built frameworks that enabled innovation — Open Finance, Pix, the RWA tokenization regime. If the legislative branch matches that precision, the outcome could be a competitive advantage, not a liability.

The critical question is who the law is written for. A framework built to address the risks of frontier foundation models — the EU's primary concern — imposes requirements irrelevant to a Brazilian HR platform using a classification model for resume screening or a Recife fintech using AI for credit decisions. Those are different risk profiles and deserve different regulatory treatments. If the Câmara doesn't make that distinction explicitly, the framework that passes will work for FANG subsidiaries in São Paulo and make life harder for the 1,400 companies building from zero everywhere else.