AI Regulation Tracker  /  Regulation in force

Brazil bars machine-only court decisions and puts every court under one national AI-oversight rule

Brazil's judiciary regulator, the CNJ, now requires human review of any AI used in the courts and gave tribunals up to 12 months to comply. A 2026 amendment ties that AI governance directly to data protection.

By Anthony Guerriero  ·  Reviewed by The Leveraged Years Editorial Desk  ·  Published July 7, 2026  ·  Last updated July 7, 2026

Brazil has adopted a single national AI rule for its entire court system, an approach the United States has not taken at the federal level. On March 11, 2025, the Conselho Nacional de Justiça (CNJ), the body that oversees Brazil's judiciary, published Resolução nº 615. It requires that any AI used in the courts operate under human supervision and forbids a decision reached by machine alone. On March 25, 2026, the CNJ amended it with Resolução nº 674, wiring the framework's governance to data protection.

The core rule is plain. Resolução 615 states that human participation and supervision are required at all stages of an AI system's use in the judiciary, and it prohibits systems that do not permit human review of their results. In practice that means a judge, or a qualified person acting under judicial authority, must remain able to inspect, correct, and reject what a tool produces. An automated output is a draft or an aid, never the ruling itself.

One national rule, not a patchwork

The reach here is what sets Brazil apart. The resolution binds all courts subject to the CNJ, which is effectively the entire national judiciary, from state civil courts to the labor and federal systems. Courts were given up to 12 months to bring existing AI projects into line, and the resolution itself took effect 120 days after publication. There is no opt-out for a court that had already deployed a favored tool. Every tribunal answers to the same standard.

To run that standard, Article 15 of Resolução 615 created the Comitê Nacional de Inteligência Artificial do Judiciário (CNIAJ). The committee was set up with 14 voting members drawn from the CNJ, the federal and labor courts, magistrates' associations, the Ordem dos Advogados do Brasil (OAB), the Public Ministry, the Public Defender's office, and civil society. It is the oversight and auditing body for judicial AI, the place where questions about a specific tool's use are meant to land.

The 2026 amendment ties AI to data protection

Resolução 674, published March 25, 2026 and signed by Minister Edson Fachin, is a targeted change with a clear signal. It raised CNJ councilor representation on the CNIAJ from two seats to three, and it made one of those seats mandatory for the councilor responsible for the CNJ's personal-data processing. The amendment also requires at least one member from the CNJ's permanent commission on information technology and innovation.

The point of that edit is governance philosophy, not headcount. By seating the data-protection councilor on the AI committee by rule, the CNJ treats AI oversight and data protection as one problem rather than two. A court system that runs AI over case files is processing large volumes of personal and sensitive data, and Brazil's regulator decided the people governing the algorithms and the people governing the data should sit at the same table.

What it does not do

The framework governs the courts, not private litigants or their counsel directly. It does not, by itself, tell a law firm which AI products it may use, and it does not replace the separate duty lawyers face when they file fabricated citations, a matter Brazilian courts have handled through bad-faith litigation penalties. It also does not ban AI in the judiciary. The CNJ chose supervision, auditability, and human review over prohibition, betting that courts can use these tools if a person stays accountable for the result.

The cross-border read

For a US professional, the contrast is the story. In the United States, individual federal judges have issued their own standing orders on generative AI, producing a patchwork that varies courtroom by courtroom. Brazil went the other way and imposed one national judiciary-wide AI rule with a standing committee to enforce it. It is a working model of centralized court AI governance, and a preview of where a more coordinated US approach could head. Any firm operating in Brazilian courts should assume the human-review standard is the floor, not an aspiration.

Browse the full AI Regulation News tracker

Informational analysis for working professionals, not legal advice. Confirm how any rule applies to your situation with qualified counsel.