Background
The spread of information manipulation (including ‘malinformation’, ‘misinformation’, and ‘disinformation’) and hate speech during elections is particularly dangerous. Disinformation and hate speech on social media can interfere with citizens' voting rights, particularly the vulnerable communities, and put the integrity of elections at risk. This is particularly relevant in the context of the upcoming Indonesian Regional Elections that were held on 27 November 2024.
The Association for Elections and Democracy (Perludem) in its Gotong Royong Against Disinformation research (2020 and 2024) found four categories of content in social media that could have a negative impact on voting rights: first, mis/disinformation regarding voting techniques/methods; second, hate speech against citizens with disabilities and their voting rights; third, mis/disinformation regarding election regulations. The research also revealed that some voters hold misconceptions, such as believing that they can vote at any polling station with an ID card, that people with disabilities are not allowed to vote, etc. (http://perludem.org/wp-content/uploads/2022/11/Buku-Gotong-Royong-Melawan-Disinformasi-Pemilu.pdf)
Furthermore, the Indonesian Anti-Defamation Society (Mafindo) reported that the amount of election-related mis/disinformation has increased significantly. They found 2,330 cases of information manipulation (more specifically we present the results of monitoring regarding ‘malinformation’, ‘misinformation’, and ‘disinformation’) in 2023. Among these, 1,292 cases were “political information manipulation”, 645 of which were related to the 2024 General Elections. YouTube was the place where most information manipulation contents were found (44.6%), followed by Facebook (34.4%), Tiktok (9.3%), X (8%), Whatsapp (1.5%), and Instagram (1.4%).
Despite the prevalence and high risk of election-related information manipulation, the Indonesian election management bodies such as the National Election Commission (KPU) and the Election Supervisory Body (Bawaslu) have expressed to the Damai Coalition that they have limited resources for social media monitoring during the meeting with Bawaslu and KPU on 6 November 2024. Therefore, the Damai Coalition and its member organizations will take participatory action to monitor social media during the 2024 Regional Election (Pilkada).
In the 2024 Presidential Election, the GUSDURian Network developed a Gardu.net monitoring tool that connected offline and online domains. Gardu.net is a platform developed by the GUSDURian Network as part of a broader initiative to monitor and ensure the integrity of elections in Indonesia, particularly for the 2024 Presidential and Legislative Election. The platform is designed to facilitate the reporting of election misconduct or alleged violations by allowing local volunteers, referred to as "Gardu Pemilu" (Election Guards), to report incidents directly from various regions. These volunteers gather information and upload it to the platform. Gardu.net functions similarly to a "community watch" system, where local "Gardu" act as the first line of defense against electoral misconduct. The platform's primary goals are to monitor elections, provide political education, and engage civil society to protect the democratic process. Additionally, it serves as a tool for civil society to engage in active participation during the election process. In total, they identified 112 cases of misconduct or alleged violations, the majority of which (62) were related to mis/disinformation regarding the integrity of state officials/institutions.
During the 2024 Regional Elections (also known as “Pilkada”), the GUSDURian Network carried out a social media monitoring initiative across three regions (East Java, South Kalimantan, and South Sulawesi) from November 19–27 to identify content related to hate speech, information manipulation (‘malinformation’, ‘misinformation’, and ‘disinformation’), offensive speech. Local observers manually monitored social media platforms, conducting targeted searches to capture region-specific content and insights that automated tools may have overlooked. The monitoring focused on major platforms, including Facebook, TikTok, X (formerly Twitter), and Instagram, which were prioritized for their substantial reach and influence in Indonesia, particularly in disseminating election-related harmful content.
Applicable freedom of expression standards
Indonesia guarantees freedom of expression through its constitution and the Law on Human Rights (Law No. 39/1999). Indonesia is also a party to the International Covenant on Civil and Political Rights (ICCPR) which guarantees freedom of expression in its Article 19. While we refer to the concepts of ‘hate speech’ and ‘disinformation’ in this report, they do not have any internationally agreed definition. Any restrictions on these types of speech must comply with the three-part test under Article 19(3) of the ICCPR.1
The falsity of information is not per se a basis for restrictions unless it is connected to one of the legitimate aims listed in the three-part test, such as the protection of public health or national security.
Based on the data obtained, several protected groups are targeted by hate speech, which can be understood in its broadest form as any expression of hate towards an individual or group defined by a protected characteristic. For hate speech, Article 20(2) of the ICCPR requires that governments restrict speech that constitutes incitement to discrimination, hostility, violence or international crimes. The UN Rabat Plan of Action, adopted in 2012 establishes a six-part test for determining whether expression rises to the level of “incitement to hatred” which may be prohibited. The six-part threshold test considers (1) context of the issues, (2) status of the speaker, (3) intention of the speaker to incite violence against a target group, (4) content and form expression, (5) extent of its dissemination, and (6) likelihood of harm, including imminence of violence and discrimination.2 In addition, the requirements of the three-part test under Article 19(3) of the ICCPR must always be fulfilled when limiting expression.
Objective
1. Identify and analyse trends related to election-related information manipulation, hate speech, and offensive speech in the context of the 2024 regional elections (Pilkada) in three regions (East Java, South Kalimantan, and South Sulawesi).
2. Produce one comprehensive report based on all monitoring findings and share it with the government and social media platforms for advocacy and awareness-raising.
Method
Conducted by teams in three regions (East Java, South Kalimantan, and South Sulawesi) from November 19–27 to identify content related to information manipulation, hate speech, and offensive speech.
Conceptual Framework
Our conceptual framework composed of three key elements along with several subcomponents:
1. Information manipulation
a. Malinformation: Information that is based on reality, used to inflict harm on a person, organization or country.
b. Disinformation: Information that is false and deliberately created to harm a person, social group, organization or country.
c. Misinformation: Information that is false, but not created with the intention of causing harm
2. Hate speech: Any kind of communication in speech, writing, or behaviour that attacks or uses pejorative or discriminatory language with reference to a person or a group on the basis of who they are, in other words, based on their religion, ethnicity, nationality, race, colour, descent, gender or other identity factor.
3. Offensive Speech: Any form of insulting speech, writing, or behavior that contains hateful content and is harmful but does not meet the criteria for hate speech, as it is not based on prejudice against a specific group identity.
Result
The GUSDURian “Gardu team” monitored social media during the 2024 Regional Election campaign from September 25 to November 23, 2024. This monitoring aimed to identify trends of information manipulation, hate speech, and offensive speech on social media to support policy advocacy and content moderation efforts.
The team conducted manual observations across various social media platforms, including X, Instagram, TikTok, and Facebook. The focus was on three provinces: East Java, South Kalimantan, and South Sulawesi. Meanwhile, Perludem monitored the regions of North Sumatra, Jakarta, and Central Java.
The analysis of the 2024 Regional Elections monitoring focused on manual tracking conducted in three regions from November 19–27, 2024, identifying content related to information manipulation, hate speech, and offensive speech.
Here are the findings and figures from the manual tracking of information manipulation, hate speech, and offensive speech on social media during the 2024 Regional Elections in three provinces:
We have identified several types of harmful content and categorized them into three sections: information manipulation which includes malinformation, disinformation, and misinformation; hate speech which encompasses racist and misogynistic content; and the most prevalent, offensive speech which includes threatening, insulting, degrading, and provoking.
This categorization helped to refine our findings of information manipulation, hate speech, and offensive speech related to the regional elections, covering a broad spectrum. During the data collection process, the “Gardu team” utilized a series of keywords associated with regional elections in their respective areas. These keywords included the names of candidates running for positions such as regent, mayor, governor, and their deputies, which were obtained from the official website of the General Election Commission (KPU). Additional keywords were formulated to capture election-related discourse, such as “Pilbup Probolinggo” (Probolinggo Regency Election), “Pilgub Kalsel” (South Kalimantan Governor Election), and “Pilwalkot Palopo” (Palopo City Election), representing three provinces and their respective municipalities or regencies. Furthermore, the team incorporated hate-related keywords often observed in narratives during previous regional or national elections, including terms like “foreign stooge” and “communist,” to ensure comprehensive monitoring of potential harmful content.
As of the report’s publication, Gardu team identified a total of 93 instances of harmful content related to the regional elections across the three regions. These included 38 cases of information manipulation, 26 cases of hate speech, and 29 cases of offensive speech. The search results across platforms revealed a distribution of 83 findings on TikTok, 5 on X (formerly Twitter), 3 on Facebook, and 2 on Instagram. However, these variations in prevalence among platforms cannot be considered representative for comparison. This limitation arises from the uneven mapping process, which did not involve equivalent sampling across all platforms.
The highest number of cases was found in East Java (54 cases), followed by South Kalimantan (29 cases) and South Sulawesi (10 cases). The most common types of content related disinformation and offensive speech, including insults and provocations against targeted candidates. Meanwhile, we identified three primordial identities that are targeted by hate speech, namely gender-based (misogynistic), ethnic-based (racist), and religious-based (bigotry). Among these three categories of hate speech, racist content (14 cases) and misogynistic content (10 cases) show a significant difference when compared to bigotry (2 cases).
In our findings, the majority of information manipulation identified in social media posts that targeted candidates took the form of content related disinformation, totaling 29 instances. This disinformation included accusations of corruption, allegations of being a communist collaborator, infidelity, involvement in “illicit” businesses, and other similar claims. For instance, on platform X, there was a post accusing several governor candidates, including the East Java governor, of being “communists” and “foreign lackeys”. We classified this type of content as disinformation. This content contains elements of hatred, but the language used is more focused on baseless accusations, particularly given that the Communist Party in Indonesia no longer exists and is banned, and there is no call to action in the content. We identified only 9 instances of malinformation, most of which were posts that cited or featured information from news articles, but were used to frame a negative narrative that could damage the reputation of a targeted candidate.
Information Manipulation, Hate Speech, and Offensive Speech Trends Related to Regional Elections in Three Regions
East Java
Candidacy Delegitimization
Based on the data collected by the “Gardu team”, East Java was the most active area in terms of information manipulation related to this year’s regional elections compared to the other two provinces. The most prevalent information manipulation in this region is the spread of disinformation. Out of the 54 instances of harmful content in East Java, nearly half were instances of disinformation (20 cases). Disinformation in East Java also represents the largest finding among the different categories of harmful content across the three regions we studied. This finding reflects the high level of political dynamics, which has the potential to divide society.
The “Gardu Team” identified “communist accusations” and “corruption allegations” against certain candidates as the two most widely disseminated forms of disinformation.
An example of this can be observed in the statement, “Jangan pilih calon gubernur/bupati/walikota yang didukung Jokowi (oligarki, antek asing dan aseng China komunis)” (Do not vote for governor/regent/mayor candidates supported by Jokowi (oligarchs, foreign agents, and Chinese Communist cronies),” posted by the account @MahesaMuna196 on the platform X on November 22, 2024. In the image he posted, the face of one of the East Java gubernatorial candidates is seen alongside a candidate from another region.
An example of another communist narrative can be seen in a video clip featuring a religious figure saying, “Said Abdullah dan bupati Sumenep, PKI! PKI! Jangan didukung!” (Said Abdullah and the Sumenep regent, PKI! PKI! Don't support them!). This video was posted by the account @faham.025 (Prabowo Madura) on the TikTok platform on November 2, 2024.
Meanwhile, corruption allegations against the candidates can be seen from several posts like the ones below:
The content posted by the account @sampangbangkit on the TikTok platform on September 26, 2024, accuses one of the Sampang regent candidates and their running mate of being former corruptors.
In addition, the account @kinzmac wrote a comment, "Sing kiri wes jelas A1 gerbong korupt dana hibah DPRD prov" (The person on the left is clearly part of the group involved in the grant funds of the Provincial Regional People's Representative Council) on the Radar Blitar media account on the TikTok platform on September 1, 2024. This account accused one of the Blitar city mayoral candidates of being involved in corruption related to grant funds handled by Provincial Legislative Council members.
As an effort to delegitimize certain candidates, accusations of being communist and involved in corruption cases are the most frequently used narratives to damage someone's reputation in Indonesia. This is quite effective given that communism is highly disliked in society due to its association with the 1965 genocide in Indonesia. Additionally, the state has legally prohibited the practice of communist ideology through the TAP MPRS Nomor 25/1966 or Decree of the Provisional People's Consultative Assembly No. 25/1966.
On the other hand, the corruption narrative is also effective in damaging the image of an official, given that the Indonesian public is sensitive to corruption cases due to the sharp economic disparity. This makes the public view corruption as one of the obstacles to their well-being.4 In addition to corruption allegations and communist issues, some disinformation content also uses narratives such as foreign agents, affair scandals, practicing witchcraft, and involvement in criminal activities. Almost all of these narratives are aimed at delegitimizing the candidates.
Targeted hatred based on gender and ethnic group
On the other hand, the number of hate speech cases (9 cases) is relatively lower compared to offensive speech (23 cases). Misogynistic content in East Java (3 cases) is also lower than in South Kalimantan (7 cases). This difference may be attributed to several factors, including the fact that all gubernatorial candidates in East Java are women, while South Kalimantan has two candidates—one male and one female.
The presence of misogynistic content that demeans female candidates based on their gender identity indicates that instances of hate speech are also present in East Java. One example is a comment made by an account on the X platform, identified as @afafufu_, which referred to one of the East Java gubernatorial candidates as a "Mulyono prostitute." The term "Mulyono" refers to the childhood name of President Joko Widodo, who has politically supported the candidate in the East Jawa gubernatorial election.
On the other hand, ethnic-based hatred also colors social media in East Java. Among the various instances of hate speech identified in East Java, racism ranks the highest in terms of frequency, with a total of four reported cases.
This ethnic-based hatred is evident in several comments on a TikTok video featuring a debate among candidates for the Blitar regency head. In the video, an account named @aviator.org wrote in Javanese, “Logate uduk Blitar” (The accent is not from Blitar). This comment received responses from other accounts, including @pak_nii, who speculated that the candidate in question was of Madurese descent, and @feronika, who wrote, “Blitar is being colonized.” These online comments reflect the presence of ethnic sentiment toward certain groups, particularly those not native to Blitar, where the majority of residents are of Javanese ethnicity.
South Kalimantan
In South Kalimantan, the most common negative trend on social media was disinformation (9 cases). The issues raised in these information manipulation were often related to business topics such as mining, discotheques, parking rates, and corruption.
Additionally, misogynistic content was quite prevalent on the region’s social media timeline during the lead-up to the regional elections (7 cases). The number of misogynistic content in South Kalimantan is the highest compared to the other two regions. The most common narrative involves the prohibition of choosing female leaders based on religious grounds and the denigration of women as leaders. Some examples of content that degrade women’s position based on their gender identity can be seen in the following examples:
Someone with the account @JenglotCrispi on the X platform wrote a comment, “Pemimpin kok binian” (a Banjar phrase meaning "How can a leader be a woman") to criticize the candidacy of one of the female candidates.
A similar comment also appeared on another platform. Someone with the account @jumiati4091 on TikTok wrote, “...sebaik apa pun seorang pemimpin itu lebih baik emang laki-laki...” (...no matter how good a leader is, it’s always better if it’s a man...) directed at the same gubernatorial candidate. This comment implies that women’s capacity will never surpass men’s, particularly in terms of leadership.
Such misogynistic comments not only target one of the gubernatorial candidates in South Kalimantan but also the candidate for deputy mayor of Banjarmasin, who is the only female candidate among the other two, as shown in the post below:
The comment written by the account @r.hidayat8 on TikTok stated that one should not choose female leaders because it contradicts religious teachings. This anti-female leadership content uses religion as a justification to position women as second-class beings in terms of leadership.
In South Kalimantan, the use of religion or statements from religious leaders advising against choosing female leaders is commonly found on social media platforms. This means that many people in South Kalimantan still blindly accept religious teachings or statements from religious leaders about matters outside the religious context.
The widespread misogynistic content found in South Kalimantan is inseparable from religious conservatism and the patriarchal culture that has developed in society. South Kalimantan has become the region with the highest number of misogynistic content because this regional election has found its moment, where several female candidates are running for leadership positions and directly facing male candidates.
Often, religion and politics are mixed together, supported by patriarchal culture. This can be seen in the same pattern of rejection against female leaders at almost every level in Indonesia, even at the presidential level.5
South Sulawesi
Meanwhile, in South Sulawesi, no instances of disinformation or misogynistic content were found. There were only ten findings, with the most common being racist content (6 cases), followed by misinformation (2 cases), and offensive speech (2 cases).
The racist content we found often reflects antipathy towards candidates perceived as outsiders or “not a son of the region”. An example of this can be found in the statement, “Ngapain pilih orang Gorontalo.. kalau ada sekampung” (Why vote for someone from Gorontalo… when there’s a local candidate?) posted by the account @pangeranmotornnk on the TikTok platform.
"Orang Gorontalo" can refer to an ethnic group, a city, or a province located in the northern part of Sulawesi. The comment indicates a tendency to prioritize candidates from South Sulawesi over those from "outside," or in this context, candidates from Gorontalo. This comment alienates one of the candidates based on their identity as someone from Gorontalo rather than being a native resident of South Sulawesi. In harsher narratives, someone might say that South Kalimantan will be colonized by people from Gorontalo, as shown in the post below:
A TikTok account named @ladoraka wrote a sentence in the Bugis language that translates to: “The husband runs for governor, the wife runs for mayor. Oh, shameless. Do you people of South Sulawesi want to be colonized by the Gorontalo ethnic group, who, in their own region, are not even chosen? Greedy. #SaveSouthSulawesi.”
Almost all findings of racist content in South Sulawesi are related to the election of gubernatorial candidates. This is because the two candidates competing there have prominent ethnic identities. One candidate carries the surname "Andi" which originates from the Bugis tribe in South Sulawesi, while the other candidate has the surname "Pomanto", which comes from the Minahasa tribe in North Sulawesi.
However, we did not find any disinformation content in South Sulawesi. This indicates that, based on the available data, South Sulawesi may have a lower occurrence of misleading information compared to other regions, particularly East Java. However, due to the lack of comprehensive data, it is not possible to conclusively determine whether South Sulawesi is more resistant to disinformation.
Overall, the findings of the Gardu team indicated that the dynamics of information manipulation on social media during the 2024 Regional Elections were greatly influenced by the social, cultural, and political characteristics of each region. East Java recorded the highest information manipulation, dominated by disinformation and insult, indicating a high intensity of political competition in the region.
South Kalimantan, on the other hand, stands out for its misogynistic issues, reflecting ongoing challenges related to gender equality. Although South Sulawesi had fewer findings, it still shows the potential for sensitive issues such as racism.
Recommendation
Information manipulation and hateful content pose potential risks to democratic integrity and social harmony. Based on our findings, we offer the following recommendations for Bawaslu and social media platforms to proactively address these issues and support a fair electoral process:
1. The findings are critical for advocating stronger policies, promoting digital literacy education, and improving regulations related to political campaigns on social media. Enhanced regulations can ensure a more transparent and democratic electoral process.
2. These findings highlight the need for collaboration with social media platforms to accelerate responses to reports of information manipulation, especially during election periods. Such collaboration can improve detection and mitigation efforts for information manipulation, hate speech and offensive speech, while observed trends can inform strategies to address hateful behavior and improve democratic practices.
3. Social media platforms have a critical role in safeguarding democratic integrity. They should improve content moderation through advanced AI tools and human oversight to detect and limit election-related manipulation information and inflammatory rhetoric. Platforms should also introduce dedicated reporting features for users during election periods, allowing them to flag specific violations, such as disinformation and identity-based hate speech.
4. Bawaslu should enhance systematic oversight of digital campaigns by partnering with local civil society organizations (CSOs). These partnerships can help monitor and address the spread of information manipulation, hate speech, and offensive speech more effectively, especially within regional contexts.
5. Public education is vital for countering negative narratives. Bawaslu should involve religious and community leaders to address misinformation, reduce polarization, and foster understanding in regions most affected by harmful content.