The Brazilian Congress's approach to disinformation and platform regulation puts freedom of expression at risk

The COVID-19 pandemic and the political crisis that followed in Brazil sparked legislative efforts to address issues related to disinformation and platform regulation.

The first drafts of the bill on “freedom, liability, and transparency online” sought to compel platforms to ban “inauthentic accounts” and “unlabeled” bots (“computer programs created to imitate, replace, or facilitate human activities in the performance of repetitive tasks on apps”—which are “not reported as such to the app provider or its users”—disinformation botnets (“a set of bots whose activity is coordinated and directed by a person or group of people, an individual account, a government, or a company in order to artificially influence the distribution of content for financial and/or political gain”), and other types of disinformation-related behavior and content. The idea of removing platform immunity under certain conditions is somewhat similar to what Trump demanded of the Federal Communications Commission in his May 28 executive order. In addition, the initial drafts threatened encryption by requiring platforms to take action against content shared in instant messaging apps. They also contained very open-ended definitions—including a definition of “disinformation”—which could lead to legal uncertainty and the risk of excessive content removals. At one point, even the creation of reputation rankings—which would affect the visibility of content—was discussed. Several other drafts were proposed by different members of Congress, and a variety of regulatory approaches are now under debate. Some of them introduce transparency obligations for platforms and free speech safeguards in content moderation, which may be beneficial. However, the proposed bills also address sensitive topics that deserve attention: Incompatibility with the intermediary liability regime set forth in the Brazilian Civil Rights Framework for the Internet (MCI): Most drafts of the disinformation bill require platforms to ban anonymous content, “inauthentic accounts,” “unlabeled bots,” and other types of content associated with disinformation, imposing heavy fines for noncompliance. Under the MCI, as a general rule, intermediaries may be held liable for content posted by their users only if they fail to remove such content after being notified of a court order requesting its removal. Risks for encryption and traceability in instant messaging apps: One of the most pressing issues at the moment is content traceability—that is, the ability to identify the origin (the author) of content shared between users of instant messaging apps. Some of the draft bills require intermediaries to store this data for up to one year. While this may be technically feasible, it would result in massive data collection from users, posing risks to their privacy. This information could be used for political persecution. Additionally, it violates the confidentiality of journalistic sources. User identification: One of the main issues raising concerns is the requirement to identify users before allowing them to create a profile on social media. This would create significant difficulties for the use of pseudonyms online and for people who use social names, such as members of the trans community. Several drafts require platforms to collect user IDs, as well as proof of address, before allowing users to create an account. A new version of the bill requires users to disclose their cell phone numbers and mandates that platforms suspend accounts linked to numbers deactivated by telecommunications companies. _ Criminalizing speech_: The most recent drafts of the bill introduce various types of criminal offenses—including new versions of crimes against honor and conceptual offenses involving defamation, disinformation, and hate speech. The definitions of the new criminal offenses under discussion are too broad and pose risks to users who could potentially be criminalized for sharing disinformation without malicious intent. Moreover, a new draft creates a new electoral offense—“sharing manipulated political advertising to disparage or ridicule candidates”—which could subject the beneficiary candidate to a heavy fine and even to removal or disqualification from the electoral race for office. In addition to all the technical issues that could jeopardize users’ rights, the main flaw in this legislative effort is the exclusion of relevant stakeholders—such as civil society, academia, and industry—from the process; everything is moving quickly, with very little room for discussion. To follow this discussion more closely, follow InternetLab on social media (@internetlabbr on Twitter, InternetLab on Facebook, and internetlab on Instagram). A new version of the bill may be voted on in the Senate next Thursday, June 25, 2020—again with very little time and space for discussion. If you wish to contribute to the debate and help raise awareness about this issue, please bring it to the attention of relevant stakeholders. Thiago Dias Oliva, head of research on Freedom of Expression at InternetLab and researcher for the Legislative Observatory @CELEUP; Heloisa Massaro, head of research on Information and Politics at InternetLab. “Photo Credit: Nijwam Swargiary”