On June 17, 2026, Brazil's Supreme Federal Court (STF) closed the last procedural chapter on one of the most consequential platform governance decisions in Latin American history. Ruling on motions for clarification in Themes 987 and 533, the Court refined — but did not retreat from — its earlier determination that Article 19 of the Marco Civil da Internet (Law No. 12,965/2014) is partially unconstitutional. Platforms now have 60 days to comply with a fault-based duty-of-care model that replaces a safe harbor that had stood for twelve years.
What Article 19 Was — and Why the Court Struck It Down
Article 19 of the Marco Civil was once celebrated globally as a model safe harbor. It insulated internet application providers from civil liability for user-generated content unless they had ignored a specific court order mandating removal. The design was deliberately conservative: it asked not whether content was harmful, but whether a court had spoken. For twelve years, this notice-and-court-order regime functioned as the backbone of Brazil's digital economy, shielding platforms from litigation costs that have chilled innovation in jurisdictions with weaker protections.
The STF's objection, articulated across Themes 987 and 533, is that this design failed to keep pace with the digital ecosystem. The Court found that requiring a court order before liability attached left platforms structurally insulated from empirically demonstrable harms — child exploitation material, incitement to political violence, mass-coordinated disinformation — regardless of how plainly illegal the content was. Critics of Article 19 have long argued in good faith that a provision built for a nascent social web in 2014 was poorly suited to the algorithmic amplification systems of 2025. That argument is not unreasonable: imposing the entire burden of adjudication on an overstretched judiciary, rather than on entities with greater technical capacity to act, carries real costs.
The Three-Tier Architecture
The replacement framework divides platform obligations across three content categories, each with distinct obligations:
Tier 1 — Evidently illicit content: Child sexual abuse material, incitement to political violence, terrorism, racism, and content undermining democratic institutions must be removed proactively, without any prior notification. Platforms that fail to detect and remove such content face automatic liability.
Tier 2 — Individually harmful content: Violations of privacy and reputation trigger a notice-and-takedown regime via extrajudicial notice — no court order required. The June 17 clarification adds that upon receipt of such notice, the platform and the content's author become jointly and severally liable. Critically, however, the ruling introduces a "reasonable doubt" safeguard: if the unlawfulness of the content is genuinely ambiguous, demonstrated uncertainty can rebut the presumption of fault.
Tier 3 — Paid and algorithmically amplified content: For sponsored posts and content boosted through recommendation systems, platforms face a rebuttable presumption of fault — they are presumed to have knowledge of content they chose to monetize or amplify.
The June 17 clarification also replaced the original language of "presumption of liability" — a near-absolute standard — with "rebuttable presumption of fault." That single phrase shift is significant: it preserves a genuine good-faith defense and signals that the Court is not building a strict-liability regime.
The Innovation Risk
The history of platform liability reform is littered with over-broad instruments that nominally targeted illegal content but induced blanket over-removal of legal speech. FOSTA-SESTA's documented chilling effect on lawful platforms in the United States and the EU's original Copyright Directive Article 13 experience both demonstrate that duty-of-care frameworks are highly sensitive to how "evidently illicit" gets defined in practice.
Brazil's three-tier model builds in several safeguards that distinguish it from blunter instruments. The narrow enumeration of Tier 1 categories — limited to specific criminal content types rather than diffuse "harmful" speech — reduces scope creep risk. The reasonable-doubt defense for Tier 2 gives platforms a genuine path to contest ambiguous removal demands. And the STF explicitly called for a "profícuo diálogo institucional" — a fruitful institutional dialogue — with the legislature, signaling that its constitutional floor is not the ceiling.
The 60-day compliance window is nonetheless tight. Platforms must establish Brazilian legal representatives, maintain accessible complaint channels, and publish transparency reports — requirements carrying real costs for smaller and mid-tier platforms without Brazil-specific legal infrastructure.
Two Decrees and a Legislative Invitation
The clarification ruling lands alongside two executive decrees — Decree No. 12,975/2026 and Decree No. 12,976/2026 — issued to operationalize the ruling's procedural requirements around transparency, notification handling, and local representation. These decrees answer the how; the deeper question of how courts will apply duty-of-care standards across edge cases remains open until case law develops or Congress acts.
Brazil's Congress has circled a comprehensive platform accountability bill for years. The STF has now handed legislators a constitutional framework and a clear invitation to fill in the details proportionately. Whether that invitation is accepted with restraint — or used to expand the liability surface beyond what the Court contemplated — will determine whether Brazil's new framework becomes a regional model for calibrated platform governance or a cautionary tale about well-intentioned reform.