EFF, Access Now, and Data Privacy Brasil are putting forward recommendations to strengthen robust privacy and data protection safeguards in the context of Brazil's elections. The recommendations stress the close relationship between violations of personal data protection and challenges to the integrity of electoral processes. They underscore how privacy and data protection guarantees are a crucial tool for curbing the targeted spread of false or manipulative content and other problematic strategies used by political actors that are amplified by digital technologies such as artificial intelligence systems.
The recommendations are part of a broader regional initiative and build on the legal and institutional safeguards already in place in Brazil. They seek to promote greater coordination among oversight institutions, civil society, and digital platforms, and encourage the solid implementation of privacy and data protection guarantees as drivers of electoral integrity. Read the full document below.
Protecting the integrity of the electoral process in the face of internet and social media use is a challenge that many policymakers are addressing or are willing to address. Online, content that can affect the integrity of the electoral process is increasingly personalized. This phenomenon is so concerning that it has been identified as one of the main global short- and medium-term risks.
In an era of generative AI, the economic cost and technical difficulty of producing and spreading false or synthetic content to deceive, manipulate, or simulate authenticity have been considerably reduced. That intensifies concern over the integrity of the electoral process. Meanwhile, online privacy and personal data protection remain unfinished business in Latin America.
There is an intrinsic connection between the ability to collect and process large amounts of personal data and the way false or manipulative content is created and distributed—on social media and messaging apps in particular, and on the internet in general. For this reason, applying strict laws and policies on personal data protection and privacy makes it possible to reduce the impact of false or manipulative content. This is especially important in electoral contexts, where such content affects and impoverishes public debate, directly affecting political and electoral rights and the integrity of the electoral process.
This phenomenon predates the emergence of the internet. However, the rise of new technologies accelerates the generation and spread of false and manipulative content. This is supported by the very economic model that sustains the platforms, amplifying its effectiveness and reach. On the one hand, social media platforms have content recommendation algorithms that use personal data to generate profiles to which they can then serve targeted advertising content, including explicitly political propaganda. This technique is known as "microtargeting."
Political microtargeting seeks to have a direct or indirect impact on democracy. It is used to persuade voters, to encourage or discourage turnout at the polls, or to raise funds using information that is deliberately taken out of context, inaccurate, or erroneous.
The control exercised by these companies raises serious concerns about people's rights. By having access to massive amounts of personal information, these companies have the ability to shape the content that users see and interact with. This happens through the construction of profiles that can reveal habits, social relationships, political preferences, and opinions, to mention a few examples. Personal data is the fuel that amplifies risks to the integrity of the electoral process. That’s true whether it’s provided by the users themselves or generated by the platforms from their interactions online.
For disinformation actors, access to sophisticated tools—such as those used to create "deepfakes" through generative AI, or "bots" programmed to spread content and seek to manipulate public opinion—boosts the effectiveness of this microtargeting in terms of quality and scalability, making it harder to detect as false or manipulative content. AI-generated avatars and synthetic characters that simulate voters, influencers, hosts, commentators, or community leaders can produce footage that appears spontaneous, fabricate the voices of artificial political actors, and make it harder for users to identify if a given public statement was created or mediated by technology.
In this context, paid promotion with nanotargeting seeks to reach increasingly specific profiles with customized content, and AI-based tools are used to assess and map its impact on social networks. Drawing on the personal data of groups of voters, profiles of "synthetic voters" are created to test messages or strategies in search of the most efficient way to influence real voters.
This rapid expansion of AI systems and hyper-personalization with data can lead to a problem of "epistemic erosion" for democratic societies, as pointed out by the UN's Independent Scientific Panel on AI Governance in 2026.
At Access Now, Data Privacy Brasil, and the Electronic Frontier Foundation, we point to the enforcement of personal data protection laws and public privacy policies as an efficient mechanism for improving the quality of our democracies and reducing the manipulation of public discourse in digital environments and its impact in electoral contexts. Measures to broaden access to information for electoral decision-making, and to ensure transparency about campaigns' and political parties' use of digital technologies built on the massive processing of personal data, also play a relevant role in guaranteeing the integrity of the electoral process.
Concern about the effects of spreading false, manipulative, or deliberately decontextualized content is particularly heightened in electoral contexts. From Argentina to Mexico, many countries in Latin America, including Brazil, are holding or will hold significant electoral processes in the coming period.
Providing the public with quality information from a range of sources is an essential element for the exercise of political rights. In order to safeguard the electoral process, these countries must enforce their privacy and personal data protection laws through their competent authorities, in coordination with their judiciaries and electoral courts.
Access Now, Data Privacy Brasil, and the Electronic Frontier Foundation propose the following recommendations to protect the integrity of the electoral process by guaranteeing privacy and data protection during electoral contexts:
1. Strengthen personal data protection guarantees and policies as a key element for the integrity of the electoral process, in particular the principles of necessity, purpose, and proportionality:
2. Political parties, federations, and coalitions must improve the information made available to the general public about their personal data processing activities in electoral contexts, including:
3. Strengthen cooperation mechanisms between the National Data Protection Authority (ANPD) and the Superior Electoral Court in order to:
4. Authorities, political parties, communicators, and social media platforms must ensure, as far as possible, that the population has access to adequate and relevant information for electoral decision-making.
5. Platforms must disable microtargeting tools for political and electoral content during previously established periods.
6. Authorities, technical actors, academics, civil society, and/or social media platforms must collaborate in creating an algorithmic impact analysis lab that makes it possible to oversee compliance with these recommendations.
7. The authorities responsible for overseeing personal data protection and electoral matters must have sufficient functional, economic, and technical autonomy and independence to guarantee the proper exercise of their powers.
Brazil's new internet intermediary liability regime is underway. The implementation of changes established by the Supreme Court includes notice and takedown mechanisms and duty of care obligations. Caution is crucial as these measures can create problematic incentives for enforcement overreach and over censorship of protected speech.
The court in June issued a new decision clarifying elements of its 2025 finding that the previous liability regime was partially unconstitutional. The government also published in late May two presidential decrees that detail how the new rules apply.
Under the new regime, social media platforms and other internet applications that curate or interfere with posts can be held liable for third-party content if they don’t remove it after being notified by the user seeking take down unless there's a reasonable doubt that the content is unlawful. For certain specific cases, like crimes against honor (e.g. defamation), platform liability still depends on failing to comply with a judicial order.
For some serious crimes, like human trafficking and crimes against women, applications have a duty of care to remove related content immediately and can be held liable when systemically failing to do so. The precise limits of what constitutes a systemic failure are still unclear. There are also stricter rules for paid ads, boosted content, and bots.
The previous regime, set by Article 19 of the law known as the Brazilian Civil Rights Framework for the Internet (“Marco Civil da Internet” in Portuguese), sought to protect freedom of expression online by holding internet application providers liable for user content if they failed to comply with a judicial order to remove it. There were specific, limited exceptions to this rule, like the unauthorized disclosure of nude or private sexual images. This was meant to prevent providers from over-removal of user content to avoid legal action. Yet, the court found that this provision failed to sufficiently safeguard democracy and fundamental rights.
We outlined the thorny context leading to this shift in Brazil’s intermediary liability rules, including Big Tech’s alignment with the far right and hurdles to approve platform regulation in Congress, through a proper legislative process.
Brazil’s shift is part of broader discussions and changes in response to growing concerns over online harms and digital platforms’ abuses. However, responses focused on platforms’ liability of user-generated content carry important traps and risks—from entrenching dominant platforms’ power over the information flow to escalating arbitrary online surveillance and censorship. The path ahead must prevent this to the extent possible, and the new presidential decrees provide a mixed contribution towards this task.
The government published two decrees regulating the new regime set by the Supreme Court. One introduces changes to its previous regulation, the Decree 8.771/2016, detailing elements of the decision, including additional duties that the court only briefly addressed (Decree 12.975). The other regulates measures to tackle violence against women online (Decree 12.976).
The Supreme Court's decision didn't establish guidelines to protect users' due process rights when facing content take down and removal demands. Instead, it relies on providers to self regulate, which could lead to over censorship.
The decrees’ provisions on user notification systems are helpful in this sense. They stipulate that providers must inform users (both the notifier and the content author) about the decision to remove or keep the content up, why, and the means to appeal. The guidance makes explicit that a platform may reconsider and reinstate content after an appeal and must explain its reasons to the party requesting removal and content author. The decrees also address concerns with the weaponization of notification systems, establishing that internet applications must adopt measures to prevent abuses.
Decree 12.795 reinforces that applications can keep content up after notification when there’s reasonable doubt that the post is unlawful, stating that the analysis should consider the context of the publications, freedom of religion and belief, and any informational, educational, or critical, satirical, or parodic purpose with the aim of ensuring freedom of expression. With these guidelines, it aims to mirror the Digital Services Act's "notice-and-action" approach. Moreover, for sexual related, intimate content, platforms will provide a specific and easily accessible notice channel where victims or their representatives can follow the case.
One of the most concerning provisions requires applications to proactively report content related to criminal conduct on their platforms to government authorities. Applications must send the post along with information that can identify the user. The Ministry of Justice will regulate this provision, something the Supreme Court didn't touch on in its decision. While it seems to apply just to those providers already required to comply with new content-related obligations (exempting email and videoconference providers, for example), it takes a disastrous step beyond. It’s not only about preventing the spread of unlawful content online; it gets platforms to police and report users to authorities by handing identification information apparently without a court order.
Decree 12.795 also details the definition of messaging applications that are exempt from notice and duty of care obligations. It excludes features for public dissemination of content and open groups so that the exemption doesn't apply. It's still unclear what exactly open groups mean. Especially regarding end-to-end encrypted applications, it's crucial that duties to monitor and take down don't affect conversations that are under this security architecture. Perhaps more troubling, the Supreme Court stated in its clarification ruling that a judicial order can determine email, voice and video conference, and messaging providers to take down content of private communications. Any measure must respect privacy and free expression safeguards and refrain from undermining end-to-end encryption.
Furthermore, decree 12.976 importantly addresses the protection of women online, but it contains a broad definition of online violence against women that will guide how platforms handle takedown notices they receive. This definition involves "any act, conduct, or omission that causes (...) psychological, political, or economic suffering (…) in any aspect of their lives, committed, instigated, facilitated, or aggravated, in whole or in part, by the use of digital technologies." Its breadth could unfortunately result in censoring legitimate criticism and other protected speech, which platforms and authorities must avoid.
The decrees also establish powers to the Brazilian Data Protection Agency (ANPD) to oversee and regulate the new regime. Among controversies, the decrees give ANPD the power to apply penalties for breaches of content-related obligations. These obligations go beyond agency competencies set in the Data Protection Law and the Law 15.211/2025, focused on the online protection of children and adolescents. They are also not clearly covered by Article 12 of Marco Civil as it stipulates administrative penalties for violations of its data privacy provisions.
We appreciate that ANPD has been open to civil society's demands and concerns. While it’s crucial that the agency conducts its oversight role preserving a proportionality commitment and keeping solid participation channels, sanction powers must be prescribed by law.
It’s true that there are critical platform accountability problems we must address, especially regarding the big players. And yes, platforms should align their policies and practices with human rights standards, including by dealing diligently with the dissemination of unlawful, toxic content. But accountability efforts should look at platforms’ systems and processes and promote measures to put checks on the power of tech giants, instead of having a prevalent focus on policing and reporting user behavior.
Key digital competition measures to regulate gatekeeper platforms are under discussion in bill 4675/2025, but the proposal is pending in Congress with no clear timeline for approval.
One important measure is to ensure accountability of take-down requests, including by the government. The Supreme Court’s decision stipulated that internet applications should publish transparency reports of the removal notices they receive. Government institutions should follow suit by periodically disclosing aggregate data of their own requests to online platforms, covering various types of user data and demands for content and account restrictions. Back in 2016, Marco Civil’s regulation decree established that all federal bodies must annually publish statistical reports on their requests of subscriber data to providers. To the best of our knowledge, federal bodies generally fail to meet this provision. ANPD can play a crucial role in stepping up transparency in the implementation of the new rules.
Ultimately, platform accountability under the new liability regime hinges on how accountable its application will be by platforms and state institutions, and on the regime's commitment to protecting fundamental rights, including freedom of expression and privacy.
EFF and ARTICLE 19 have submitted joint comments to the European Commission on draft guidelines for the Digital Services Act’s trusted flagger mechanism. Having long advocated for a DSA that protects freedom of expression while preserving intermediary liability protections and the prohibition on general monitoring, we welcome the Commission's effort to provide practical guidance on how the trusted flagger system should operate.
The DSA’s trusted flagger system can help platforms identify illegal content more efficiently. But if implemented poorly, it could also encourage over-removal of lawful speech, weaken due process, and give government authorities disproportionate influence over online expression.
We support the Commission's focus on good practices and illustrative examples, rather than legal interpretations that could inadvertently steer platforms toward particular enforcement outcomes—and argue that the guidelines should include stronger safeguards to protect freedom of expression, due process, and the impartiality of the trusted flagger system.
We also support the Commission's clarification that the DSA itself does not define "illegal content"; that determination must come from applicable national or EU law. Trusted flaggers submit prioritized notice, but platforms remain responsible for determining whether content is actually illegal. Platforms must therefore conduct careful, informed assessments and should not assume that a trusted flagger notice necessarily warrants restricting content.
Our submission highlights several areas where the guidelines could be strengthened:
Read the full submission here:
Users are always seeking more control over their social networking experience to make it better, whether to improve privacy or enhance flexibility. Interoperability between social networking platforms like Facebook and TikTok has so many benefits that solve those issues.
Say you’re on multiple platforms because you have friends you follow on different networks, but you’ve decided to choose one platform with better privacy practices. With interoperability, you could switch and still interact with friends who remain on larger platforms. It could also enable independent apps with better privacy controls and more user choice. These are the untapped possibilities that could benefit users in the European Union under the 2022 Digital Markets Act (DMA).
Yet, the European Commission, in its first review of the DMA, announced in April it had decided not to extend the DMA’s interoperability mandate to social networking and didn’t give a deadline or a timeline for enforcing that part of the Act. The Commission said “there is no clear demand” from users and businesses for social networking interoperability and, in any case, it’s too technically complex at the moment. Meanwhile, the Big Tech platforms that have been slow-walking interoperability over the last two years, erecting a myriad of hurdles for users seeking more freedom to choose other platforms, get a pass.
This is a huge disappointment and a missed opportunity by the Commission. Interoperability dismantles one of the biggest barriers faced by users who want to leave the tech giants’ platforms: the choice between changing to a platform you prefer or staying behind on a platform where all your friends, communities, and customers are.
The DMA, which went into force in 2024, aims to foster more choices for European Union users and encourage competition and innovation by forcing so-called gatekeeper platforms like Meta, Apple, and Google, to open their ecosystems to competitors. The regulation does a great deal to foster the integration of competing services and devices with the ecosystems of very large online platforms that act as gatekeepers. It even requires interoperability for messaging services, despite the significant technical and privacy challenges involved.
So, it’s odd that the Commission is using complexity as a shield against taking on social networking interoperability. The internet already runs on complex interoperable systems. Approaches like ActivityPub, the decentralized networking protocol behind the “Fediverse,” which gave rise to decentralized networks like Mastodon, already exist. The DMA shouldn’t mandate a specific protocol, but it can require meaningful interoperability outcomes.
The argument that there’s no real demand for social networking interoperability also falls flat. Users want the ability to move across platforms, choose the content they’d like to see from platforms, and not be tied down to a single platform. But there’s no way to get there—the platforms are doing little to open their social networking ecosystems. And now you have the DMA’s enforcer saying it’s not going to make them change. Demand for alternatives won’t materialize at scale until users see real progress towards interoperability, something the Commission has the power to do.
Having decided there’s little demand and too much complexity to proceed with mandating social networking interoperability, the Commission said it “will continue to monitor and assess how these services evolve.” This wait-and-see-posture only hurts users and strengthens and further entrenches Big Tech incumbents.
The DMA is supposed to center on the rights of technology users and be the pathway to an internet experience where you decide which software runs on your devices, where it’s easy to find the best products and services, and where you can leave a platform for a better one without forfeiting your social relationships.
Meanwhile, Big Tech is also resisting the DMA’s openness requirements. For example, Apple is supposed to be opening up iOS devices to rival app stores. Yet, the smartphone giant’s plan for opening its App Store levies junk fees and onerous conditions on app makers and is effectively impossible for any competitor to use.
It’s not just Apple pushing back against DMA enforcement. Meta's response is a “pay for privacy “system, in which users who do not consent to Meta’s surveillance will have to pay to use the service, or be blocked from it. Whether their plan complies with the DMA remains under review.
Nowhere in the DMA does it say social networking companies get to install a toll booth for users seeking to benefit from privacy rights the regulation grants them. The future EU Digital Fairness Act is another opportunity to protect users from such practices by declaring them unfair.
The Commission has responded to these developments with investigations, preliminary rulings, and fines. Meanwhile, users are missing out on greater choice and flexibility in how they communicate and connect online.
Seeking transparency and accountability in Paraguay’s use of facial recognition, EFF, the Association of Technology, Education, Development, Research, Communication (TEDIC), and the Centre for Justice and International Law (CEJIL) filed a complaint with the Inter-American Commission on Human Rights against the state for arbitrarily denying access to information about its implementation and use of the technology as a tool for mass surveillance that erodes people’s privacy rights.
The case involves the Ministry of the Interior and National Police’s installation in 2019 of surveillance cameras with facial recognition technology in Asunción. Maricarmen Sequera, a lawyer and executive director of TEDIC, filed an information request with the ministry seeking details and protocols about the implementation and use of facial recognition systems and the personal data processing involved.
The request sought information about, among other things, whether the state had conducted human rights or data protection impact assessments, as well as if it had developed measures and protocols for avoiding abuses, illicit uses of personal data, and other risks in the deployment of the facial recognition system.
The state denied most of the information requested, arguing that implementation details, protocols, and the processing of individuals' personal data were confidential security information. TEDIC contested the secrecy in courts, but the analyses lagged and ultimately sustained the denial of information.
The petition filed last Friday (19) cites Inter-American standards upholding the public’s right to access information, particularly in relation to national security, that the Paraguayan authorities disregarded in denying TEDIC’s information request. The petition also argues that the refusal of information violated privacy and the right to informational self-determination.
The petition asks the Commission to recognize a violation of those rights and require the state to deliver the information requested. Further, the petition seeks an order compelling the state to adopt mandatory permanent mechanisms of active transparency regarding the acquisition, contracting, implementation, financing, functioning, and use of surveillance technologies by public bodies, especially those that incorporate processing of biometric data or artificial intelligence systems.
It also asks the Commission to order the state to mandatory procedures for human rights impact assessments prior to acquiring and using surveillance technologies, particularly those that collect biometric data or use artificial intelligence.
The state’s lack of transparency in this case is not an isolated incident, both in Paraguay and in Latin America, where opacity in matters of security and surveillance is the unsettling rule. The situation gets worse with the increasing normalization of intrusive surveillance technologies by states in the region.
The Special Rapporteur for Freedom of Expression of the Inter-American Commission emphasized that states should disclose surveillance capabilities and contracts, and acknowledge state use of surveillance technologies at a meaningful level of detail, to facilitate essential public debate on the necessary limitations of surveillance in democratic societies and ensure compliance with international human rights law.
We hope that the Inter-American Commission upholds the robust safeguards in the Inter-American System and advances access to information and privacy rights in a case that can set a crucial precedent for the region.
Login