Brazil may soon become the first country in the Americas to regulate digital platforms before they do anything wrong. Bill 4,675/2025 would let the Administrative Council for Economic Defense (CADE) designate large technology companies as having “systemic relevance” and impose special obligations without first proving anticompetitive conduct or consumer harm.
That is a substantial change in how competition law works. The bill would allow CADE to regulate self-preferencing, default settings, interoperability, and data use in advance rather than through ordinary case-by-case enforcement.
By the government’s estimate, five to 10 companies would be designated, most of them likely American. The stakes therefore extend beyond competition policy. Foreign regulation of U.S. technology firms has become a trade and foreign-policy flashpoint, and Brazil’s proposal could add another point of friction.
The bill also tests a broader claim now shaping digital regulation around the world. Can traditional antitrust enforcement police dominant platforms, or should governments supplement it with pre-emptive rules modeled on the European Union’s Digital Markets Act (DMA)?
Brazil offers a revealing test. Its competition authority has recently shown what existing law can accomplish. Settlements with Apple and Google produced DMA-style outcomes, including greater openness in Apple’s iOS operating system, through ordinary case-by-case enforcement. Early evidence from Europe, meanwhile, increasingly points to consumer harm and weaker incentives to innovate rather than the promised surge in competition.
Brazil’s choice could influence other emerging economies weighing similar regimes. That gives the debate in Brasília significance well beyond Brazil.
In “Digital Overreach: A Premature Turn to Ex Ante Regulation in Brazil,” Geoffrey Manne, Dirk Auer, and I argued that Brazil neither needs nor would benefit from a new ex ante regime for digital markets. The narrower question here is whether the substitute text now before the Chamber of Deputies improves on the government’s original proposal and how seriously it addresses the criticisms advanced by the International Center for Law & Economics (ICLE) and others.
The substitute adopts many of the procedural and institutional safeguards we recommended. It leaves the proposal’s doctrinal core intact. That core remains the strongest argument against enacting the bill.
Fast-Tracked, Then Stalled
The executive branch submitted Bill 4,675/2025 in September 2025. The measure would amend Brazil’s Competition Law, Law 12,529/2011, to let CADE designate large technology platforms as having “systemic relevance” in digital markets and impose “special obligations” on them.
Two recent procedural developments pushed the proposal closer to a vote.
The first came March 18, when the Chamber approved an urgency motion, REQ 4612/2025. Under the Chamber’s rules, urgency allows a bill to bypass the standing committees and proceed directly to the plenary floor. That compresses the process into negotiations among party leaders and the bill’s rapporteur, stripping away the procedural checkpoints where legislation is usually tested, amended, delayed, or quietly buried.
The second came July 8, when the rapporteur, Congressman Aliel Machado of Paraná’s Green Party, filed his report and a substitute text after roughly three months of review. Party leaders had agreed that the report would clear the way for a floor vote, and the Chamber’s website now lists the bill as ready for the agenda.
The substitute matters because it is, for practical purposes, the text the Chamber will consider. Members may still offer amendments on the floor, but under urgency, the rapporteur’s draft becomes the baseline for every negotiation that follows.
The timing remains uncertain. An attempt to bring the bill to a vote during the week Machado filed his report failed to win consensus among party leaders. On July 15, the Chamber held its final deliberative session before the winter recess and again left Bill 4,675 off the agenda.
That leaves a narrow window for action this year. Once lawmakers return to their states, the Chamber is expected to concentrate its legislative work before Brazil’s general elections in October and November 2026. Some expect a vote during the first week of August, after the recess. Brazilian politics advises against putting that date in ink.
Even Chamber approval would settle only half the matter. Brazil has a bicameral Congress, so the bill would then move to the Senate, where the debate would begin again.
New Guardrails, Same Bureaucracy
The substitute gives the bill’s new enforcement unit a longer name. It would now become the “Special Superintendence for Systemic Relevance, Free Competition and Consumer Protection in Digital Markets.” The change has no direct legal effect, but it fits Machado’s claim that the bill “regulates markets, not companies.”
The more consequential revisions limit administrative rulemaking. Article 18-A bars the resolution defining the superintendent’s duties from creating new sanctioning powers, expanding the grounds for designation, or imposing obligations without a statutory basis. CADE’s complementary regulations would be limited to implementing the deadlines and procedures established by law. They could not create additional obligations, sanctions, or grounds for designation.
That change responds directly to a concern running through ICLE’s analysis. The original bill paired open-ended designation criteria with broad administrative discretion. Brazil’s institutional constraints make that combination especially risky, even if similar powers appear in some of the foreign regimes the bill seeks to emulate.
The substitute leaves the institutional structure intact. ICLE recommended placing the digital-markets function within CADE’s existing General Superintendence, following the approach used by the European Commission, the United Kingdom’s Competition and Markets Authority (CMA), and Germany’s Bundeskartellamt. Instead, the bill retains a parallel superintendence that duplicates CADE’s architecture and creates room for jurisdictional conflict.
Worse, the substitute provides no new appropriations to fund it. The bill therefore preserves both the institutional fragmentation and the resource constraints that ICLE identified in the original proposal.
More Choice, Less Discipline
Article 47-B sets out the new regime’s objectives. The original bill listed three goals: reducing barriers to entry, protecting the competitive process, and promoting freedom of choice. The substitute keeps all three and simply moves “freedom of choice” to the front of the line.
That reshuffling does not cure the underlying problem. In the “Digital Overreach” white paper and an earlier Truth on the Market post, we argued that Article 47-B departs from the consumer-welfare framework embedded in Brazil’s Competition Law. None of its three objectives contains a limiting principle tied to measurable effects on consumers.
Barriers to entry may reflect scale economies, quality investments, or network effects that make products better or cheaper. “Protecting the competitive process,” as Herbert Hovenkamp has put it, is little more than a slogan. It is circular, vague, and difficult to measure.
“Freedom of choice” may be the weakest standard of all. Joshua Wright and Douglas Ginsburg have shown that a choice-based test detaches antitrust from consumers’ actual preferences and the tradeoffs among price, quality, innovation, and variety. It treats more options as inherently better, even when fewer options produce lower prices or better products.
CADE’s own decisional record points in the same direction. “Freedom of choice” appears only once as a proxy for welfare. Price appears 94 times.
We recommended making consumer welfare the regime’s explicit objective and limiting principle. The bill’s three stated goals could then serve, at most, as indicators of possible harm. The substitute declines that recommendation.
That omission creates a deeper statutory tension. The bill would amend Brazil’s Competition Law rather than establish a separate regime. Article 36, Section 1, of that law recognizes that dominance achieved through efficiency and the “natural process” of competition is lawful. Article 88, Section 6, conditions merger approval on benefits reaching consumers.
An ex ante regime centered on “freedom of choice” fits awkwardly within that framework. Sooner or later, CADE will have to decide which principle governs.
A Better Gate, the Wrong Key
The designation provisions receive the substitute’s most consequential revisions. These rules determine which companies CADE may classify as having “systemic relevance in digital markets,” Brazil’s term for digital gatekeepers. In an interview, the rapporteur acknowledged that the original test was, to some degree, open-ended.
The original bill allowed CADE to designate a company based on seven listed characteristics considered “non-cumulatively” and “among others.” The list was therefore nonexhaustive, and any one factor could, in theory, justify designation. The company also had to exceed either R$50 billion in annual global revenue, roughly US$10 billion, or R$5 billion in Brazilian revenue, roughly US$1 billion.
The substitute replaces that framework with a two-step, closed test. The revenue thresholds become a threshold condition. CADE must then conduct a “joint and reasoned analysis” (análise conjunta e fundamentada) of six listed characteristics. The phrase “among others” is gone.
The substitute also changes how the thresholds may be adjusted. The original allowed ministers to revise them by joint act. The new text indexes them annually to Brazil’s IPCA inflation measure. That removes a political dial and prevents inflation from quietly pulling more firms into the regime, as has happened with Brazil’s merger-review thresholds.
A new provision also makes explicit what the original bill left unclear. Designation alone would not impose any special obligation. Each obligation would require a separate process and justification.
Most of these revisions move in the direction ICLE recommended. The “Digital Overreach” white paper criticized the original framework for combining open-ended criteria, politically adjustable thresholds, and a lengthy designation period. The substitute closes the list, fixes the threshold adjustment, requires a reasoned assessment of multiple factors, and, as discussed below, shortens the designation period.
Two problems remain. First, the six characteristics are still linked by “or,” even though the statute requires a “joint and reasoned” analysis. The drafting points in two directions at once.
Second, some supporters argue that a higher bar would weaken the regime. That objection ignores the cost of designation. A designated firm, along with its entire corporate group, could face years of regulatory obligations, compliance reports, and heightened sanctions risk.
An error-cost approach weighs the consequences of mistaken intervention against mistaken restraint. In fast-moving digital markets, false positives can be especially costly and hard to reverse. A designation carrying such consequences should therefore require substantial evidence.
The deeper problem remains unchanged. The test still does not require proof of market power.
The rapporteur’s official release says the bill “does not start from the premise that large companies are a problem.” The statutory test suggests otherwise. ICLE recommended requiring proof of durable market power, consistent with the United Kingdom’s Digital Markets, Competition and Consumers Act (DMCC), which requires “substantial and entrenched market power,” and Germany’s Section 19a, which requires a finding of dominance.
The substitute instead retains the “systemic relevance” framework. Its six factors remain structural proxies, including multisided business models, network effects, vertical integration, and access to data. Those features may reflect competitive success rather than durable dominance.
The test also omits the questions a competition economist would ask first. Can new firms enter? Are new technologies disrupting the market, including artificial intelligence? Can users switch to substitutes?
Designation still applies to the entire corporate group, although the substitute’s service-specific obligations reduce some of the practical breadth. The government also continues to estimate that five to 10 firms would qualify, most of them U.S.-based. That prospect carries obvious geopolitical costs and suggests that company size, rather than demonstrated consumer harm, remains the test’s center of gravity.
On its own terms, the substitute improves the designation test. It is narrower, more predictable, and more demanding. Its central conceptual flaw nonetheless survives.
Fewer Blunt Instruments, More Fine Print
Article 47-E lists the special obligations CADE may impose on designated firms. The substitute reorganizes them into three categories: transparency and reporting duties, positive obligations, and abstention obligations. The changes are substantive, and they cut both ways.
The clearest improvement concerns merger review. The original bill required designated firms to submit every transaction to CADE, regardless of whether it met the turnover thresholds in Article 88 of Brazil’s Competition Law. That would have created a parallel merger-control regime.
We argued that the rule was disproportionate. It would capture harmless deals, threaten the startup exit path on which venture investment often depends, and duplicate authority CADE already has under Article 88, Section 7, to call in below-threshold transactions. CADE has recently shown that it will use that power in AI acqui-hire investigations.
The substitute scales the rule back. Designated firms must now notify the superintendence of below-threshold transactions, following the approach in Article 14 of the DMA. Any review would proceed through CADE’s existing call-in authority. The notice requirement remains blanket rather than risk-based, but the bill drops its most burdensome feature.
The second improvement is mandatory tailoring. Each special obligation must apply only to specified services, products, or business practices identified in the proceeding. CADE must calibrate the obligation to “the competitive risks associated with each product or service,” and the Tribunal must identify the covered services and set an implementation timetable.
That change partially adopts ICLE’s recommendation that remedies attach to particular services rather than entire corporate groups. It should prevent the kind of companywide obligations that have made the DMA so costly to administer. The corporate group would still receive the designation, but each obligation would target a particular digital product or service.
The substitute also strengthens the balancing clause. The original text said CADE “may consider” information security, a firm’s other legal duties, and the functioning of the relevant product or service. The substitute says CADE “shall consider” those factors, as ICLE recommended.
It also adds two new considerations: innovation, especially for small firms, and the “public interest in the competitive development of digital markets.” The latter phrase sheds little light. Still, mandatory consideration gives CADE a stronger duty to explain how it weighed competing concerns.
The larger omission remains. The bill still provides no efficiencies defense, a problem discussed below.
Against those gains, the substitute adds several new mandates. One provision would allow CADE to require “neutral and non-discriminatory” choice architecture, including choice screens, for selecting, installing, or setting third-party products as defaults. Dirk Auer and I have explained to the UK CMA why such remedies can raise consumers’ search costs without producing measurable gains in competition.
A new abstention rule would also bar firms from worsening conditions for users who exercise rights created by the special obligations. That includes “subversion of user autonomy” through interface design, an anti-circumvention and anti-dark-patterns provision.
The remaining changes are mixed. The vague ban on “predatory or abusive strategies” survives, though it now applies only where a firm “exploits the situation of dependence” of users. The anti-steering rule grows broader, extending to restrictions “of any other nature.”
That expansion ignores the role some anti-steering rules play in protecting platform investments in product discovery, trust, and security. CADE’s settlement with Apple showed that those tradeoffs can be assessed through case-by-case enforcement.
The self-preferencing ban also remains unchanged. Economic research does not support treating self-preferencing as presumptively harmful, and CADE has never based a digital-market conviction on it.
More Process, Less Ambush
The procedural revisions work best as a package. Together, they make the regime slower, more participatory, and harder to rush. They also respond to ICLE’s concerns about legal certainty, institutional capacity, and rent-seeking.
The designation period drops from as long as 10 years to six. A firm may seek review after two years if market conditions change significantly. Special obligations may also be revised after two years, and they would take effect within 90 days rather than 60.
ICLE recommended designation terms of three to five years with mandatory periodic review. Six years, with review triggered only by changed conditions, still exceeds the five-year limits under the UK’s DMCC and Germany’s Section 19a. It is nonetheless a marked improvement over 10 years.
Participation also expands. CADE must open a 30-day public-comment period when a designation proceeding begins. The consultation period doubles from 15 to 30 days, and the final opinion must expressly address the submissions received.
A new voluntary-proposal process under Article 87-C allows a designated firm to submit implementation plans, technical parameters, monitoring mechanisms, and timetables. The proposals would not bind CADE, but the agency would have to consider them and explain its response.
That is a modest step toward the negotiated, case-specific approach ICLE has argued better suits digital markets. CADE has limited experience prescribing product-design remedies. A rule developed with technical input from the firm will often work better than a rigid order that engineers must retrofit after the fact.
The changes are not uniformly positive. “Any interested party” may still participate, and several public bodies can force CADE to open a proceeding. Referrals from CADE’s Tribunal, its General Superintendence, the Finance Ministry’s Secretariat for Economic Reforms, or any federal body responsible for digital markets or “diffuse and collective rights” trigger immediate proceedings and automatic intervenor status.
The superintendence may now dismiss unsupported private complaints, which helps. But the bill still needs stronger safeguards against firms and advocacy groups seeking regulatory favors at a rival’s expense.
The Alternative Becomes an Add-On
The substitute also creates two institutions absent from the original bill. Article 87-K establishes a nonpunitive market-study process. CADE could publish reports to inform designation decisions and evaluate, ex post, whether existing special obligations worked as intended.
The bill also creates a nonbinding Advisory Council. At least half its seats would go to academics and nonprofit civil-society representatives, with appointment rules meant to reduce the risk of capture.
ICLE recommended market studies as an alternative to an ex ante regime. The substitute instead adds them to one. Both institutions are sensible on their own. Neither answers the objections to the regime they would support.
Better Guardrails, Same Destination
The substitute adopts many of the recommendations in the “Digital Overreach” white paper. It closes the designation list, requires reasoned analysis, shortens the designation period to six years, makes clear that designation does not automatically trigger obligations, and requires CADE to tailor each obligation to specific services. It also strengthens the balancing clause, reduces the proposed parallel merger regime to a notice requirement, limits rulemaking, and expands public participation.
The rapporteur plainly read the criticism and responded to much of it. Four of our central recommendations nonetheless remain unresolved.
First, the bill still lacks a consumer-welfare anchor. Article 47-B rearranges its objectives but does not reform them. Nothing requires CADE to justify a designation or obligation by showing expected benefits to consumers.
Second, the bill still lacks a market-power standard. “Systemic relevance” continues to turn largely on size and structural characteristics rather than proven, durable power in a defined market.
Third, the bill still offers no efficiencies defense. Requiring CADE to consider certain factors improves the procedure, but it does not give a designated firm the right to defeat or narrow an obligation by showing that the challenged conduct produces benefits that outweigh any harm.
That defense appears elsewhere in Brazilian competition law. Articles 36, Section 1, and 88, Section 6, of Brazil’s Competition Law recognize efficiencies as legally relevant. The new regime does not. As a result, the same conduct could be lawful under the statute’s general provisions because it benefits consumers, yet prohibited under the digital-markets regime because efficiency provides no defense. The substitute leaves “economic justification” undefined, so the inconsistency remains.
Fourth, the bill still requires no regulatory impact assessment, even though Brazil’s Economic Freedom Act, Law 13,874/2019, requires such analysis for regulatory measures of far less consequence.
The substitute therefore produces two conclusions. It is a serious effort to add guardrails, and on several procedural dimensions, Brazil’s proposal is now more constrained than the DMA.
Those guardrails also expose the central defects more clearly. CADE could still designate a firm without proving durable market power. It could still impose a tailored obligation that the firm cannot challenge by showing that the conduct benefits consumers.
A well-run process is no substitute for a sound legal standard. Better machinery only makes the missing metric harder to ignore.
A Global Trend, or Just a Rerun?
The rapporteur has also pointed to developments abroad as evidence that the international trend still favors ex ante regulation. Beyond the European Union, United Kingdom, and German regimes already in force, he cites the June 10 reintroduction of the American Innovation and Choice Online Act (AICOA) by Sens. Amy Klobuchar (D-Minn.) and Chuck Grassley (R-Iowa).
Context matters. This is AICOA’s third appearance after failing in the 117th and 118th Congresses. As my colleague Daniel Gilman put it, the latest version “is not so much a fresh start as a sequel nobody ordered.” It arrived with four original co-sponsors, in an election year, and with no clear path that its predecessors lacked.
Reintroduction alone does not amount to regulatory momentum. It is a thin basis for claiming that Brazil would be joining a global movement rather than racing ahead of one.
Nor does AICOA offer a model worth copying. Geoffrey Manne observed that the new version “fails to fix the bill’s central legal and economic flaws—and in several ways makes them worse.” It still identifies covered firms, now called “systemically important platforms,” through formulas based on revenue and user reach, without requiring proof of market power.
Herbert Hovenkamp’s critique tracks the central problem with Article 47-C. AICOA selects firms “on the basis of raw size rather than market power,” moving competition law away from anticompetitive conduct and its effects and toward structural traits that may have no connection to consumer harm.
The same criticism applies to Brazil’s bill. Brazilian law already recognizes the contrary principle. Under Article 36, Section 1, of Brazil’s Competition Law, a company does not violate antitrust law merely by becoming large or dominant through efficiency.
The rapporteur has shown that the bill can be drafted more carefully. The substitute adds real procedural safeguards, and further revisions could improve it again.
But better drafting cannot answer the threshold question. Brazil still has not shown why it needs this regime in the first place.
A sharper scalpel does not justify an unnecessary surgery.
