Big techs reduce election measures and face new judicial rules in Brazil

To a greater and lesser extent, Meta, X and Google have been aligning themselves with the Trump administration's banner of reducing interference in publications circulating on social networks and video platforms. In Brazil, there was also a reduction in specific measures for the election.

Here, however, big techs will have to comply with new legal requirements in combating electoral disinformation. Companies will have the challenge of balancing respect for local laws and the new direction coming from the USA, after what they consider to have been excesses in content moderation during the Covid pandemic and the attacks against the Capitol in 2021.

TikTok, of Chinese origin, is the only platform that is going against the grain - increasing its bet on content and program moderation for the Brazilian election.

In 2022, the main "regulation" of digital political content was the pen of the then president of the TSE (Superior Electoral Court) Alexandre de Moraes, ordering content to be removed, in addition to the court's resolution shortly after the first round that tightened the measures and deadlines for removing disinformative content.

In 2024, in the municipal elections, the propaganda resolution reported by minister Carmen Lúcia, then vice-president of the TSE, instituted the requirement for labels for content made with artificial intelligence and banned deepfakes. But, in that election, contrary to what was predicted, there was no flood of AI disinformation.

This year, a scenario is emerging in which AI disinformation will be much more prevalent. Additionally, there are new rules. The propaganda resolution published in March introduces a total ban on political content with AI in the 72 hours before and 24 hours after the election. The text prohibited artificial intelligence chatbots such as ChatGPT, Gemini and Grok from recommending votes or favoring candidates in user queries.

For the first time, platforms are required to present compliance plans detailing how they intend to comply with the TSE resolution with goals and achievement metrics and reports documenting the results. Companies have until August 16th to submit their plans to the TSE.

Anticipating a record increase in judicialization, one of the big techs, for example, increased by 30% the number of external lawyers who will work in elections to deal with requests for content removal and other judicial determinations.

At the beginning of the year, executives from this big tech projected a 30% increase in the number of lawsuits compared to 2022 - now, the estimate is high between 50% and 60%.

Furthermore, there was the decision of the STF that changed the Marco Civil da Internet, increasing the responsibility of companies for user content, and the Executive decree that designates the ANPD (National Data Protection Agency) as the regulatory authority for networks. The changes do not directly address electoral content, but may touch on it in some cases.

At the same time, internet platforms will have fewer measures aimed at combating electoral disinformation in Brazil than in 2022.

"There are two vectors, in opposite directions. On the one hand, in the last two years some platforms have adopted a more liberal approach to speech, which has created conditions for less content moderation. On the other, in the last year Brazil has instituted rules that create incentives for more intervention. The synthesis of these two movements is what will be at stake during the election", says Fernando Gallo, managing partner of the consultancy Input.

In January 2025, just before Trump's inauguration in the US, Meta founder Mark Zuckerberg announced that the company would "get rid" of fact-checking. The measure has not yet been implemented in Brazil, where Meta maintains a partnership with six checking agencies. However, it reduced the number of publications checked.

According to the report, Meta cut the maximum volume of agency checks by around 20%, which results in a reduction in verified content. Every election year, the company usually increases the goal of checked content, but has not yet done so this year.

Meta also discontinued some of the programs to combat electoral disinformation that it had in 2022. Posts about elections no longer had labels with links that directed TSE information. According to the company, it was assessed that these labels lost their effectiveness over time. This year, there will be no chatbot on Instagram for users to access electoral information. And the company has not published an updated version of the women in politics guide, unlike in 2022 and 2024.

But the Megafone tool was maintained, which provides voters with reminders from official electoral sources before the election, such as dates for registering to vote and information about polling locations.

Unlike 2022, Brazil has not been designated a temporary high-risk location. But, according to Meta, this can happen, if deemed necessary.

To curb deepfakes, Meta provides tools for users to self-declare AI-generated content. "During the campaign period and the legally mandated AI quiet period, additional content generation restrictions, detection and enforcement measures may be implemented," the company says.

Meta, like all other companies, signed a memorandum of understanding with the TSE with generic measures to combat electoral disinformation. In their terms of use, they all commit to removing content that incites violence, misinforms about voting dates and locations or leads to vote suppression.

At Google, there were two changes to specific measures for elections. In March 2022, YouTube announced that it would begin removing content alleging fraud in the 2018 election.

This year, however,