Brazilian Superior Electoral Court Approves New Resolution on Artificial Intelligence and Freedom of Speech

On February 27, 2024, the Brazilian Superior Electoral Court (“TSE”) – responsible for judging electoral cases and overseeing the organization of elections - approved 12 resolutions that will guide the 2024 municipal elections, including Resolution No. 23,610, dated December 18, 2019 ("Resolution")[1].

The Resolution marks a notable evolution in the regulation of the Brazilian electoral landscape, addressing a wide range of topics from ensuring responsible use of artificial intelligence (“AI”) to safeguarding freedom of expression and controlling misinformation AI.

Its main target is to ensure an ethical and transparent use of AI, thereby protecting the fairness of the democratic process. TSE defined “Artificial Intelligence” as a: 

“computational system developed based on logic, knowledge representation or machine learning, obtaining an architecture that enables it to use input data from machines or human beings to, with a greater or lesser degree of autonomy, produce synthetic content, predictions, recommendations or decisions that meet a set of previously defined objectives and are capable of influencing virtual or real environments.” 

Moreover, it defined “synthetic content” as an “image, video, audio, text or virtual object generated or significantly modified by digital technology, including artificial intelligence”.

Among the measures outlined in the Resolution concerning this issue, we highlight the following:

1.Implementation of tools to identify and combat the spread of misinformation and fake news;

2. Requirement of labels for the identification of synthetic multimedia content generated with AI to create, replace, omit, merge or change speed or superimpose images or sounds. The labels shall be adequate to the medium where the ad was broadcasted (e.g., at the beginning of radio ads or through watermarks in campaign posters);

3. Restrictions on the use of chatbots and avatars simulating interaction with a candidate or other real person; and

4. Absolute prohibition of the use of deep fake (i.e., digitally generated or manipulated content, even with authorization, to create, replace, or alter the image or voice of a living, deceased, or fictitious person cannot be used to harm or favor a candidacy).

TSE has also provided that the identification labels are not required in cases in which the AI is used to:

1. improve image or sound quality;
2. generate graphic elements of visual identity, vignettes and logos; and
3. generate materials that were already common in past elections (e.g., posters with images of allied candidates that were digitally manipulated to simulate that they took the photo together)

Among other prominent aspects of the norm, the protection of the freedom of expression of artists and influencers stands out. This measure is essential for preserving the plurality of ideas and a healthy democratic environment for political debates. By acknowledging the crucial role these agents have in shaping public opinion, the Resolution seeks to ensure that their creativity is not unduly limited during electoral periods. Therefore, artist and influencers may share their political position during shows, performances, and on the internet, as long as it is a deliberate manifestation, free of any payment.

In addition to protecting freedom of expression, the Resolution aims to ensure the protection of the personality rights of artists and creators of audiovisual works. Considering the intrinsic value of these works and their personal connection to the authors, the norm establishes mechanisms to prevent the unauthorized use of these works in electoral campaigns. According to the TSE, injunctive relief, provided as an appropriate mean to cease unauthorized use, represents an effective tool to protect the rights of creators and ensure that their artistic contributions are duly recognized and respected.

Regarding copyright, the Resolution prohibits the use of audiovisual works, music, jingles, and sound effects in electoral propaganda without the proper authorization from copyright holders, as a measure to secure the recognition and adequate remuneration of these creators during the electoral period.

Furthermore, it tackles an issue that has troubled musicians in recent campaigns when candidates were free to use famous songs simply changing parts of the lyrics because the alterations were enough for them to claim in court that the new song was a parody and, according to the Brazilian Copyrights Act (Federal Law No. 9,610/1998), parody is fair use. TSE’s Resolution expressly states that authors may request the cessation of ads containing parodies of their work.   

In effect, the Resolution aligns with other existing legal norms, both in the electoral context and in the realms of copyright and freedom of expression.

In the electoral context, the norm complements legislation aimed at ensuring the fairness of the electoral process and equality of opportunities among candidates.

From the perspective of information dissemination, the Resolution imposes obligations on application providers, including measures to combat the spread of misinformation and ensure swift and effective action in taking down irregular content and accounts during the electoral period. These measures aim to protect the integrity of the democratic process, preventing the spread of misinformation, hate speech, fake news, and other forms of manipulation that could negatively influence voters.

Noteworthy requirements include:

1. Implementation and disclosure of measures to prevent or reduce the spread of notoriously false or severely decontextualized information that may affect the integrity of the electoral process;
2. Promotion, through cost-free boosting, of informative content that clarifies notoriously false or severely decontextualized information;
3. Maintenance of an advertisement database for real-time monitoring of content, values, sponsors, and demographic characteristics of the population groups that make up the audience (profiling) of contracted ad; and
4. Immediate take down of content and accounts during the electoral period in cases of risk, such as:
   1. Behavior, information, and antidemocratic acts provided by the Criminal Code; and
   2. Hate speech or hateful behavior, including racism, homophobia, Nazi and fascist ideologies, or hateful actions against an individual or group based on prejudice of origin, race, sex, color, age, and any other forms of discrimination.
   3. Content manufactured or manipulated, partially or entirely, by digital technologies, including AI, in disagreement with the labeling set out in the Resolution.

However, it is important to note that this regulation is not without debates and challenges, fitting into a discussion that is not new in the Brazilian legal context.

Indeed, the Brazilian Civil Rights Framework for the Internet (also known as “Marco Civil da Internet” in Portuguese – Federal Law No. 12,965/2014), establishes fundamental principles such as network neutrality and freedom of expression. Its Article 19[2] outlines the responsibility of application providers for third-party content only in cases of non-compliance with specific court orders requiring the removal of certain content deemed illegal (e.g. hate speech and copyright infringement). 

However, this responsibility is conditioned upon the non-compliance with a specific court order, thus ensuring the preservation of freedom of expression while also protecting against abuse and legal violations.

This issue gained even more prominence with Extraordinary Appeal No. 1,037,396[3], currently under analysis by the Brazilian Supreme Court (“STF”), addressing the constitutionality of the above-mentioned Article 19 and the possibility of civil liability of internet providers, websites, and managers of social media applications for damages resulting from third-party illicit acts.

Nonetheless, TSE’s Resolution demonstrates a proactive approach by the state in regulating the digital environment during the electoral period, where providers may be held responsible if they do not act proactively to prevent damages, emphasizing the importance of their actions to ensure the integrity of the democratic process.

In this context, TSE’s Resolution represents a significant progress in specific regulatory approaches for the electoral context. Meanwhile, debates on the responsibility of digital platforms continue to be shaped by the complex interaction between constitutional principles, specific laws, and emerging concerns related to misinformation and cybersecurity.

In summary, TSE Resolution No. 23,610 represents a comprehensive effort to modernize and strengthen the Brazilian electoral system while protecting citizens' fundamental rights and promoting transparency and integrity in the democratic process. By addressing issues such as freedom of expression, misinformation, and use of AI and other new Technologies, the resolution demonstrates a clear commitment to promoting democracy and strengthening electoral institutions. 

The 2024 municipal elections are scheduled to happen on October and will be the first Brazilian election since AI has become an accessible tool. Thus, it will serve as a test for the Resolution and to TSE, which will face new challenges and will have to live up to them.

……………

[1] Available in Portuguese at: https://www.tse.jus.br/legislacao/compilada/res/2019/resolucao-no-23-610-de-18-de-dezembro-de-2019

[2] “Article 19. With the aim of ensuring freedom of expression and preventing censorship, the internet application provider may only be held civilly liable for damages arising from content generated by third parties if, after a specific judicial order, it fails to take the necessary measures, within the scope and technical limits of its service and within the stipulated period, to make the content pointed out as infringing unavailable, except as otherwise provided by law.

Paragraph 1. The judicial order referred to in the main clause must, under penalty of nullity, clearly and specifically identify the content pointed out as infringing, allowing unequivocal location of the material.

Paragraph 2. The application of the provisions of this article to infringements of copyright or related rights depends on specific legal provisions, which must respect freedom of expression and other guarantees provided for in Article 5 of the Federal Constitution.

Paragraph 3. Cases involving compensation for damages arising from content available on the internet related to honor, reputation, or personality rights, as well as the unavailability of such content by internet application providers, may be brought before special courts.

Paragraph 4. The judge, including in the procedure provided for in paragraph 3, may, based on clear evidence of the fact and considering the community's interest in making the content available on the internet, anticipate, in whole or in part, the effects of the protection sought in the initial request, provided that the requirements of the plaintiff's claim being likely and the reasonable fear of irreparable or difficult-to-repair damage are present.” The full text in Portuguese is available at: https://www.planalto.gov.br/ccivil_03/_ato2011-2014/2014/lei/l12965.htm 

[3] Available in Portuguese at: https://portal.stf.jus.br/processos/detalhe.asp?incidente=5160549