On June 17–18, 2026, Brazil's Supreme Federal Court (STF) concluded its consideration of nine motions for clarification in Extraordinary Appeals RE 1.037.396 and RE 1.057.258 — the cases behind Themes 987 and 533. The ruling did not reopen the 8–3 decision of June 26, 2025, which partially struck down Article 19 of the Marco Civil da Internet (Law 12.965/2014). It refined the implementation architecture in ways that matter enormously for every internet application provider operating in the country.
The immediate consequence: a 60-day compliance clock running from June 18, 2026, to implement duty-of-care obligations covering the immediate removal of serious unlawful content. The Court also replaced the prior 'presumption of liability' for paid and artificially amplified illegal material with a 'rebuttable presumption of fault' — a targeted legal refinement that still places the evidentiary burden on platforms to show they acted without knowledge or negligence.
Twelve Years of Safe Harbor, and Why the Court Moved
For twelve years, Article 19 of the Marco Civil da Internet functioned as Brazil's intermediary safe harbor. Under it, platforms bore civil liability for user content only after receiving a specific, individualized court order. The legislature's intent was deliberately protective: to prevent well-resourced actors — state institutions, corporations, powerful individuals — from weaponizing extrajudicial removal requests against critics and journalists.
That logic was not without merit. Any platform required to act on private complaints becomes susceptible to coordinated suppression campaigns. Court-order requirements create friction that protects minority voices and investigative reporting. Three STF justices — André Mendonça, Edson Fachin, and Nunes Marques — dissented precisely because they saw greater risk in weakening that protection than in preserving it.
The majority concluded otherwise. Serious criminal content — terrorism, child sexual abuse material, incitement to ethnic violence, anti-democratic mobilization — was circulating at scale on platforms that faced no legal consequence for inaction absent a court order. That model, the Court found by 8 votes to 3, was incompatible with Brazil's constitutional framework of fundamental rights.
What the June 2026 Ruling Actually Changed
The clarification ruling made three targeted refinements.
Rebuttable presumption of fault for sponsored and amplified content. Unlawful material in paid advertisements, sponsored posts, or content distributed through inorganic amplification mechanisms now triggers a presumption that the platform was at fault. Platforms can rebut it by demonstrating they acted without knowledge and without negligence. The shift from 'presumption of liability' to 'rebuttable presumption of fault' matters: the former approaches strict liability; the latter preserves a meaningful defence and allows proportionality arguments.
Joint and several liability upon extrajudicial notice. Once a provider receives a substantiated notice of unlawful content, it becomes jointly liable with the content author if it fails to act — no court order required for this category. A safeguard applies: a provider that documents a diligent assessment and finds genuine uncertainty about the content's unlawfulness can rebut liability. How courts define 'reasonable doubt' in this context will be contested in Brazilian litigation for years.
Executive regulatory authority confirmed. The ruling expressly affirms that the Executive Branch may issue regulations within the framework the Court established. This is not a procedural aside — it authorises presidential decrees and ministerial regulations that could operationalise platform obligations without waiting for Congress.
The three-tier content structure from June 2025 survives intact. Serious criminal offences — terrorism, child sexual abuse material, hate speech, anti-democratic conduct, incitement to self-harm — require proactive duty-of-care systems irrespective of notice. General unlawful content triggers joint liability upon extrajudicial notice. Crimes against honour — defamation, slander — still require a court order before liability attaches. That last point matters: the original Article 19's strongest free-speech protection was preserved for the category of content most frequently weaponised against journalists and political opponents.
What Remains Unresolved
Brazil is home to 150 million social media user identities and 185 million internet users. Content-moderation mandates at that scale carry real consequences, and the clarified framework still leaves critical operational terms undefined.
'Systemic failure' — the trigger for duty-of-care liability on serious criminal content — has no regulatory definition. Neither has 'adequate measures,' the standard platforms must meet to demonstrate compliance. 'Identical content,' which governs follow-on removal obligations after an initial notice, also lacks clear parameters. As the Global Network Initiative has noted, the framework applies to all internet application providers — not merely the Very Large Online Platforms targeted by the EU's Digital Services Act. A framework calibrated to Meta's operational capacity will be disqualifying for a small civil-society platform or a community-run news forum that lacks a dedicated legal team.
The STF's ruling called for 'fruitful institutional dialogue' (prófícuo diálogo institucional) with the legislature — an explicit invitation to Congress to supply the definitional work the Court could not. That invitation remains open. The 60-day deadline arrives before Congress or the Executive can plausibly define those terms. Platforms must build compliance programs against incomplete specifications, a condition that advantages incumbents over entrants and deepens structural market concentration the framework was never designed to create.
Where Brazil Sits Globally
Brazil's approach charts a course between two established poles. The United States' Section 230 grants near-blanket immunity for third-party content — a model increasingly contested but durably resistant to legislative revision. The EU's Digital Services Act is explicitly risk-tiered: the most onerous obligations fall on designated very large platforms; smaller services receive proportionately lighter treatment.
Brazil's hybrid is broader in scope than the DSA: duty-of-care obligations apply across the provider spectrum, not just to platforms above a defined user threshold. It is more nuanced than a full Section 230 repeal: the court-order requirement for honour offences — the category most sensitive to political weaponisation — is preserved. The design is coherent in principle. Whether it is workable depends on definitional clarity that the June 2026 clarification, for all its refinements, did not provide.