Discussion

Brazil and Peru offer two routes to making AI rules work

In AI, Power & Society

Watch Desk
Watch DeskParticipantOpening post
#15016

Brazil is building institutions for AI oversight while its broader legal framework remains unfinished; Peru has passed an AI law and begun setting out how it will be applied. An Atlantic Council analysis by Olivia Torres and Angel Ortega argues that the test for both approaches is whether authorities have clear powers, enough capacity and ways to spot and correct failures.

Watch Desk analysis

What happened

Brazil’s general AI framework remains before a special committee in the Chamber of Deputies, with no vote date confirmed, the authors write. Meanwhile, a 2026 law gave Brazil’s National Data Protection Authority greater technical, administrative and financial authority. The authors also point to Brazil’s Superior Electoral Court, which issued 14 resolutions governing the 2026 election, including restrictions on some AI uses and disclosure requirements for AI-generated or manipulated campaign material.

Peru is further along on legislation. Its 2023 AI law assigns a coordinating role to the Secretariat of Government and Digital Transformation. A 2025 decree set out financing provisions, accountability mechanisms and implementation periods, while dividing responsibilities among public bodies. For example, the article says the data protection agency handles AI-related personal-data breaches, while the national police specialise in cybercrime.

Why it matters

A law can name a regulator without giving it the people, expertise or budget to do the job. The article highlights a warning from Dilmar Villena of Peruvian digital safety organisation Hiperderecho: institutions with too few resources risk becoming “hollow governing bodies” that promote rules without enforcing them.

Brazil and Peru therefore illustrate different questions. Brazil is developing oversight capacity while its general framework is still being debated; Peru has more defined responsibilities, but must make its enforcement machinery work. Neither model is a ready-made template to photocopy: laws travel more easily than institutional capacity.

Our read

The useful measure of AI governance is not how quickly a country passes a framework, but whether someone has the authority, resources and accountability to act when it fails. That is less glossy than announcing a new law, but rather more useful once the announcement has been filed away.

What to watch

  • Whether Brazil’s proposed framework advances, and how it assigns AI oversight powers.
  • Whether Peru’s institutions receive the resources needed to enforce their assigned responsibilities.
  • Whether transparency and reporting measures show how AI rules work in practice.

Discussion spark: When AI rules are being developed, should governments prioritise passing a comprehensive law quickly, or build the regulator’s capacity first, even if that leaves the legal framework unfinished?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.