EXECUTIVE SUMMARY
- Malaysia’s democratic foundations face critical challenges in the digital age, with racially charged polarisation and the proliferation of disinformation during election campaigns posing significant threats to electoral integrity.
- The lack of robust, continuous, and trusted fact-checking mechanisms overseeing political information in Malaysia has exacerbated the country’s vulnerability to misinformation and disinformation, even outside of electoral campaigning seasons.
- Malaysia’s current regulatory framework is ill-equipped to handle the speed and volume of digital disinformation, with sluggish implementation of existing laws and protracted legal processes failing to provide timely recourse against political disinformation.
- The rapid advancement of artificial intelligence (AI) technologies has elevated the spread of disinformation to unprecedented levels, making it increasingly difficult to distinguish truth from falsehood and posing a serious threat to the integrity of public knowledge.
- To address these challenges, this article proposes several policy recommendations, including establishing an independent fact-checking body, investing in AI-powered fact-checking tools, developing country-specific AI governance guidelines, implementing more agile regulatory mechanisms, and introducing comprehensive media and information literacy education.
* Nuurrianti Jalli is Visiting Fellow at the Media, Technology and Society Programme at ISEAS – Yusof Ishak Institute. She is also Research Affiliate at the Data and Democracy Research Hub at Monash University, Indonesia, and Assistant Professor at the School of Media and Strategic Communications at Oklahoma State University. Khairy Jamaluddin is Associate Senior Fellow at ISEAS – Yusof Ishak Institute. Previously, he served as Minister at Malaysia’s ministries of Youth and Sports, Science and Technology, and Health. He was also the Coordinating Minister for the Covid-19 Immunisation Programme.
ISEAS Perspective 2024/72, 18 September 2024
INTRODUCTION
Malaysia’s democratic foundations face a critical challenge in the age of digital transformation. While the manipulation of information for political advantage is not a new phenomenon, it has evolved to unprecedented levels of complexity and influence. The Malaysian context provides stark examples of how unchecked allegations and false narratives can spread rapidly across digital platforms, potentially destabilising the delicate balance of the country’s multicultural society and democratic processes.
During politically charged periods, particularly around elections, false narratives and polarising propaganda have frequently fuelled divisive sentiments. This was vividly illustrated in the aftermath of the 2022 general election, when calls for a second “May 13”[1] incident began trending on TikTok following the announcement of Malaysia’s first hung parliament. The invocation of May 13 – a dark chapter in Malaysian history marked by violent clashes between ethnic Malays and Chinese in 1969[2] demonstrates how social media platforms can rapidly amplify sensitive historical memories. This is fuelled by the structure of social media platforms, which allows for rapid sharing, and a business model that benefits from promoting and amplifying sensational content.[3]
Various unfounded claims and misleading information have circulated, impacting multiple political parties and candidates. For instance, accusations of widespread voter fraud, or “undi hantu” (ghost voting), have been propagated without substantial evidence.[4] Similarly, unverified allegations of corruption against various political figures have also rapidly gained traction on social media, fuelled by trends like “Hoi Ya Hoi.” This phrase, popularised by Malaysian singer Altimet’s song “Aduh Malaysia,” centres on corruption allegations against political leaders and has potentially influenced voter perceptions without proper investigation or fact-checking. Additionally, sexist narratives have surfaced during campaigns, targeting women candidates with degrading and baseless accusations. In one example, during Malaysia’s 15 General Election in 2022 (GE2022), Muhammad Sanusi Md Nor of the Malaysian Islamic Party (PAS) went viral on TikTok and Twitter for disparaging female leaders as incompetent.[5]
These examples underscore the urgent need for robust safeguards to protect the integrity of Malaysia’s electoral processes in the digital age. The electoral system faces a clear and present danger, primarily stemming from interconnected threats: racially charged polarisation, the proliferation of disinformation during election campaigns, and the rapid spread of unchecked allegations across social media platforms.
DISINFORMATION AS A STRATEGIC CAMPAIGNING TOOL
While disinformation and propaganda circulate daily, their spread intensifies during emotionally charged events such as natural disasters, public health crises, and election periods.[6] These moments of heightened tension create fertile ground for the rapid proliferation of false or misleading content.
In Malaysian politics, the use of disinformation as a political tool is not new. The key difference between today’s landscape and that of three decades ago is the presence of the Internet. Since the early 2000s, the exploitation of information for political gain, particularly using hateful ethnoreligious sentiments and character assassination tactics on the Internet has not only intensified existing societal divisions but also further threatened Malaysia’s multicultural democracy. Now, with over 83% of Malaysians active on social media,[7] these platforms have become fertile ground for malicious actors to propagate hateful propaganda,[8] spread inflammatory content,[9] and couch extremist views in seemingly patriotic rhetoric.
GE2022 vividly illustrated this problem. On TikTok, for example, conspiracy videos labelled Chinese political candidates as communists, claiming that voting for them would be a disservice to the Malay community. Similarly, within the Chinese community, narratives suggested that voting for non-Chinese candidates could jeopardise the wellbeing of Chinese Malaysians in a Bumiputera-majority country. This rhetoric was not limited to TikTok but also appeared on other social media platforms such as Facebook, Twitter, YouTube, and WhatsApp. Research conducted by the Centre for Independent Journalism (CIJ) highlighted the scale of this issue. They found that, across four social media platforms—TikTok, Facebook, Twitter, and YouTube—the number of unique messages touching on common hate speech subjects nearly doubled to 99,563 between 20 October 2022 and November 2022, compared to 55,000 in an earlier study conducted over a longer period from 16 August 2022 to 30 September 2022.[10]
Often written in local languages and dialects, this content was hard to moderate.[11] Consequently, this allowed for hateful rhetoric to spread and provided loopholes for propagandists to exploit. As a result, during GE2022, problematic hashtags such as #BangsaMelayu, #KetuananMelayu, #13Mei, and #HoiHoiYaHoi trended on TikTok, and similar hate-generating initiatives circulated on other social media platforms.
| Hashtags | Context |
| #13Mei | This hashtag was used to call for a second racial riot in Malaysia (referencing 13 May 1969) if non-Malay individuals were to become key political figures and form the government after the election. |
| #BangsaMelayu #KetuananMelayu | These hashtags were used by Malay ultranationalists alongside #13Mei to call for a second racial clash in Malaysia if the Democratic Action Party (a Chinese-majority party) formed the government during Malaysia’s first hung parliament. |
| #HoiYaHoi | This hashtag was used to character-assassinate many political candidates during GE15. “HoiYaHoi” is part of Malaysian singer Altimet’s song (also contested during GE15) “Aduh Malaysia” which centres on corruption allegations against political leaders. |
These are among the many examples over the years of how social media platforms got weaponised to exacerbate ethnic tensions and manipulate voter sentiments in Malaysia.[12] By appealing to deep-seated fears and prejudices, such content distorts the electoral landscape, potentially swaying voters based on racial fears rather than policy considerations. This trend undeniably poses a significant threat to Malaysia’s delicate multicultural balance and the integrity of its democratic processes. The combined effect of racial polarisation and widespread disinformation is undeniably profound. These erode public trust in democratic institutions, further polarising the electorate along ethnic and religious lines. Such a toxic information ecosystem undermines the very foundations of Malaysia’s multicultural democracy.
INDEPENDENT FACT-CHECKING BODY
The absence of strong, continuous, and trusted fact-checking mechanisms in Malaysia has heightened the country’s vulnerability to misinformation and disinformation. This gap has drawn criticism from local political observers, who argue that fact-checking is ineffective, despite evidence from various studies showing that independent fact-checking agencies play an effective role in curbing the spread of false information.[13] [14] The lack of such safeguards allows false narratives to proliferate, even outside of election campaigns when public attention is less concentrated on political discourse.
Alarmingly, Malaysia’s leading fact-checking agencies, including Sebenarnya.my[15] and MyCheck,[16] when contacted, explicitly state that they do not fact-check political content. This leaves a significant void in the verification of political information, precisely where it is most needed. Smaller fact-checking initiatives tend to operate on a seasonal basis, primarily verifying information during general elections. This sporadic approach fails to address the ongoing nature of political mis- and disinformation and leaves the public vulnerable during non-election periods.
Another fact-checking initiative, JomCheck,[17] which was established in 2022 as an academe-media-civil society fact-checking alliance, represents a positive development. During GE2022, JomCheck played a role in fact-checking information. However, the impact of their efforts remains uncertain, as the initiative struggled to gain significant traction on social media, limiting public awareness of its existence. Additionally, its effectiveness is further constrained by its grant-based funding model, which may not provide the consistent resources needed for sustained and comprehensive fact-checking. Also, the involvement of the National University of Malaysia, a public institution, as one of the main founders of the alliance raises concerns about potential state influence over content evaluation. Learning from the history of state scrutiny over faculty activities in Malaysian academe, this concern is legitimate. In particular, under ACT 605 Rule 17(b), faculty and any university officers are prohibited from making public statements that “could embarrass or damage the reputation of a statutory body or the Government”.[18] This rule poses a significant risk of being breached if JomCheck were to debunk contentious political statements made by any members of the Malaysian government. Despite JomCheck’s collaborations with media agencies and civil society organisations such as Astro Awani, Malay Mail, Bersih, and the Centre for Independent Journalism, the potential for bias or self-censorship remains a significant concern. This situation highlights the delicate balance needed in creating truly independent fact-checking mechanisms.
In our conversations with fact-checkers in Indonesia and the Philippines, a significant issue contributing to the rapid spread of misinformation is raised. This is with regards to the sheer volume of such content shared every minute on social media platforms, a problem exacerbated by limited social media moderation efforts; the content is often posted in local languages with deep local nuances, making it difficult for automated systems to identify and flag. Consequently, the responsibility to debunk this misinformation falls heavily on fact-checkers.[19] However, due to these systemic limitations, including a lack of human resources and inadequate technology to handle the high volume of content, only a limited amount of misinformation can be properly fact-checked. The situation is worsened by an inadequate level of media and information literacy among voters, which makes them more susceptible to manipulation through false or misleading content. Many citizens lack the skills to critically evaluate online information.[20] This lack of critical evaluation skills not only allows misinformation to spread unchecked but also destabilises the democratic process.
SLUGGISH LAWS AND REGULATIONS
Malaysia’s current regulatory framework is also ill-equipped to handle the speed and volume of digital misinformation, particularly during critical election periods. While existing laws such as the Sedition Act 1948, Defamation Act 1957, Communication and Multimedia Act 1998, Penal Code Section 499 (governs criminal defamation) and the recently passed Cyber Security Bill 2024[21] are in place to mitigate disinformation and political falsehoods, their implementation has been frustratingly sluggish.
This sluggishness stands in stark contrast to the swift action taken by tech giants like Meta when faced with government pressure. For instance, in August 2024, Meta quickly apologised and restored posts by Malaysian Prime Minister Anwar Ibrahim related to the assassination of Hamas leader Ismail Haniyeh after initially removing them. This incident followed a similar occurrence in May, where Meta also restored Anwar’s posts about meeting Haniyeh.[22] The rapid response came after Malaysia’s communications minister, Fahmi Fadzil and members of the Prime Minister’s Office met with Meta representatives, highlighting the government’s ability to exert pressure on social media platforms.
However, when it comes to addressing disinformation and defamation cases through legal channels, the process remains excruciatingly slow. Defamation cases can drag on for years, providing ample opportunity for political candidates, parties, and extreme loyalists to propagate hateful narratives during elections without fear of immediate repercussions from authorities. An example of this systemic failure is the ongoing defamation case between former health minister Dzulkefly Ahmad and former Prime Minister Najib Razak. Filed in 2021 after Najib accused Dzulkefly of cynicism and nepotism, the case was only set to go to trial in June 2024 – a delay of over three years.
The disconnect between the rapid spread of digital disinformation and the slow pace of legal redress creates a dangerous environment where false narratives can flourish unchecked. While Meta can swiftly reverse decisions on content removal when pressured by the government, the broader legal system lacks this agility in combating widespread disinformation. The contrast is striking: Meta’s quick climb-down regarding government-related Palestine content demonstrates that rapid responses are possible in the digital sphere. Yet, the Malaysia’s legal system’s inability to match this speed in addressing political disinformation issues leaves a significant gap in protecting the integrity of public information, especially during elections.
EMERGING TECHNOLOGIES AND MISINFORMATION ON STEROIDS
The rapid advancement and availability of sophisticated tools like artificial intelligence (AI) have elevated the spread of misinformation to unprecedented levels. AI-driven technologies can produce vast amounts of highly convincing misinformation at an alarming rate, making it increasingly difficult for the average person to distinguish truth from falsehood. These tools can create deep fakes, manipulate images and videos, and generate realistic text that mimics genuine news articles or social media posts. The sheer volume of this AI-generated content threatens to overwhelm traditional fact-checking methods, a concern repeatedly highlighted by professional fact-checkers from Masyarakat Anti Fitnah (MAFINDO)[23] and Alliance of Independent Journalists (AJI)[24] in Indonesia and VERAFiles[25] in the Philippines.
Moreover, AI’s ability to micro-target audiences exacerbates the problem.[26] By analysing vast amounts of data, AI can identify and exploit individual and group vulnerabilities, tailoring disinformation to specific demographic segments.[27] This precision targeting can manipulate people’s beliefs and behaviors more effectively than broad-based propaganda efforts. In the context of elections, such micro-targeting can sway voters’ opinions, reinforce existing biases, and polarise communities. The use of AI in disseminating political disinformation is not just a technological issue but a profound societal challenge that requires urgent attention from policymakers, technology companies, and civil society.
While there is no strong indication of the widespread use of AI tech for strategic campaigning yet, the potential for its use in the near future is highly likely. Malaysia has already experienced increased political polarisation, and the infusion of AI-generated misinformation into the political landscape could accelerate this dangerous trend. This creates a vicious cycle: as misinformation spreads, trust in traditional information sources erodes, and vice versa.
POLICY RECOMMENDATIONS AND THE WAY FORWARD
To address these challenges, we propose a multi-faceted approach.
Non-partisan, independent fact-checking agency free from potential political influence
First, Malaysia must establish an independent, non-partisan fact-checking body that operates year-round, not just during elections. As mentioned earlier, affiliations with the state or any political parties could compromise the integrity of any fact-checking initiative. This independent body should therefore collaborate instead with the likes of Malaysia’s Electoral Commission (EC), civil society, and scholars. Fact-checking should cover not only social media posts but also statements made by political candidates throughout the electoral season. Working with the EC, for example, would improve trust in electoral bodies and serve as a check and balance on electoral conduct.
AI powered fact-checking tools with local LLMs
To effectively address the challenges of misinformation in Malaysia, investing in AI-powered fact-checking tools specifically designed for the Malaysian context is crucial. These tools should aim to understand and process local languages[28] and dialects to assist in identifying and addressing false information efficiently.[29] However, it is important to approach this development with a nuanced understanding of both the potential and current limitations of AI in fact-checking including data bias,[30] and potential political influence over content interpretation.[31]
While AI has shown promise in tasks such as detecting AI-generated deepfakes through metadata analysis and forensic techniques,[32] its ability to verify contentious claims in real time remains an evolving capability. Determining which sources are considered “authentic” still requires human expertise and ongoing refinement of AI systems. Developing these systems must involve data from multiple stakeholders, particularly independent political observers and media professionals, to ensure a balanced and accurate approach. The complexities of detecting false information, especially synthetic media, and the challenges in real-time claim verification highlight the need for cautious optimism about AI’s role in fact-checking. As studies have noted, AI’s effectiveness is still limited in smaller languages and nuanced contexts, and these limitations should not be overstated.
Despite these challenges, there is significant potential in further developing AI technology to enhance information verification. A study by the University of Washington has shown that fact-checkers are already leveraging AI to assist with tasks such as transcription, translation, document synthesis, assisted search, image analysis, trend analysis, and other core natural language processing (NLP) functions. However, these developments are still in their early stages, particularly in the Majority World (or Global South), where infrastructure and resources are lagging behind.
To harness this potential responsibly, it is crucial to support local startups in developing Malaysian-focused large language models (LLMs) for fact-checking purposes. These startups must operate independently of any state or political party affiliations to ensure the integrity of the tools they create. Both the government and the private sector should offer development grants aimed at creating robust AI-powered LLMs. These grants should be carefully structured to prevent political interference. Transparent and impartial oversight mechanisms should be established to oversee both the grant allocation process and the development of these critical tools.
AI governance and country specific guideline
Third, while the ASEAN Guide on AI and Governance[33] is available, country-specific AI governance regimes should still be developed to oversee the use of AI in various sectors in Malaysia. These tailored regulations would monitor the use of AI across various sectors in Malaysia, ensuring not only the ethical implementation of the technology but also safeguarding national sovereignty and the wellbeing of Malaysian citizens. Country-specific guidelines should address the unique cultural, economic, and social contexts of Malaysia, allowing for more precise and relevant governance of AI technologies. This approach would enable us to harness the benefits of AI while mitigating potential risks that may be specific to our national landscape. By developing our own AI governance framework, we can better protect fellow Malaysians’ interests, maintain control over our technological development, and ensure that AI applications align with Malaysian values and priorities. And these guidelines would serve as a crucial tool in preserving Malaysia’s digital sovereignty, reducing dependence on foreign AI systems that may not fully comprehend or prioritise our needs. This proactive stance in AI governance would position us as a leader in responsible AI adoption within Southeast Asia.
Building on the need for country-specific AI guidelines, Malaysia should implement stringent measures specifically for the use of AI in political campaigning during elections. These measures are crucial to maintain transparency and fairness in the democratic process. During election periods, the EC should issue and circulate strict instructions to all contesting parties regarding the use of AI in their campaigns. These guidelines should explicitly address the ethical use of AI technologies, with a particular focus on protecting voter privacy and preventing manipulation.
One key aspect of these guidelines should be imposing limitations on the use of user data for targeted political messaging. This would help prevent the exploitation of personal information for political gain and ensure that voters are not unduly influenced by hyper-personalised campaign tactics. Contesting parties found in violation of these regulations would face penalties under the recently amended Personal Data Protection Act (Amendment) 2024[34] which imposes fines of up to RM 1 million and/or imprisonment for up to three years. This stringent approach will underscore the seriousness of data privacy violations in the political sphere and serve as a deterrent against unethical campaign practices.
More agile regulatory mechanism
To strengthen our defences against disinformation and hate speech, we must comprehensively enhance our regulatory framework. This involves developing more agile mechanisms for content moderation, implementing stricter penalties, and accelerating the legal process for addressing the deliberate spread of false information and inflammatory rhetoric. These measures must be balanced with the protection of freedom of expression to ensure a fair and just approach.
A key aspect of this improved framework is the enhancing of social media accountability. Malaysia’s recent move to require licensing for social media services with over 8 million users[35] from 1 August, is a step in this direction. This initiative aims to ensure compliance by social media companies with Malaysian laws designed to combat cyber offenses. However, it is crucial to emphasise that this new regulation must be implemented and enforced with utmost care to avoid inadvertently restricting freedom of speech.
Social media companies should be required to implement robust, transparent content moderation systems with clear guidelines on addressing harmful content. These guidelines must be narrowly tailored to target specific harms as defined by established laws, without arbitrary application. The focus should be on preserving diverse viewpoints and fostering open dialogue while effectively combating clearly illegal content. Social media companies should publicly disclose their content moderation policies, regularly update them based on evolving threats, and provide detailed explanations when content is removed, or accounts are suspended.[36] This approach ensures a balance between protecting users from genuine harm and safeguarding the free exchange of ideas. Moreover, it serves to protect national sovereignty, ensuring that local laws and cultural norms are respected without undue influence from foreign corporate entities.
Media and information literacy in schools
While there are a small number of initiatives to be found in Malaysia which focus on media and information literacy, a more comprehensive and widespread approach is urgently needed. Introducing media and information literacy as a key subject in schools and higher education institutions is crucial for building a resilient and discerning society. This approach should be integrated across all levels of education, from primary schools to universities. The curriculum should cover critical thinking skills, digital literacy, fact-checking techniques, and understanding the mechanisms of misinformation spread, tailored to be age-appropriate to each educational level.
And, these programmes should extend beyond formal education settings to reach all segments of society, including older adults who may be more vulnerable to online misinformation. Community centres, libraries, and online platforms can offer workshops and resources to ensure widespread access to these vital skills. By fostering a culture of critical thinking and media savvy, we can empower Malaysians to navigate the complex information landscape effectively.
Increased collaboration and institutional support
Finally, fostering collaboration between government bodies, tech companies, civil society organisations, and academia is crucial in combating misinformation. This united front will create a more comprehensive and effective strategy by leveraging diverse expertise and resources. Government agencies can provide legal frameworks and enforcement, tech companies can offer technological solutions and platform-specific insights, civil society organisations can ensure public interests are protected, and academia can contribute research-based approaches. By working together, these stakeholders can develop coordinated responses to mis- and disinformation, share best practices, and adapt quickly to emerging threats in the digital information landscape.
The integrity of our electoral process is the bedrock of our democracy and protecting it in the digital age requires collective effort and vigilance. As we navigate this critical juncture, the actions we take now will determine the resilience of Malaysian democracy for generations to come. By implementing reforms, fostering a culture of critical thinking, and promoting digital responsibility, we can ensure that our elections remain free, fair, and truly representative of the will of the people.
ENDNOTES
For endnotes, please refer to the original pdf document.
| ISEAS Perspective is published electronically by: ISEAS – Yusof Ishak Institute 30 Heng Mui Keng Terrace Singapore 119614 Main Tel: (65) 6778 0955 Main Fax: (65) 6778 1735 Get Involved with ISEAS. Please click here: /support/get-involved-with-iseas/ | ISEAS – Yusof Ishak Institute accepts no responsibility for facts presented and views expressed. Responsibility rests exclusively with the individual author or authors. No part of this publication may be reproduced in any form without permission. © Copyright is held by the author or authors of each article. | Editorial Chairman: Choi Shing Kwok Editorial Advisor: Tan Chin Tiong Editorial Committee: Terence Chong, Cassey Lee, Norshahril Saat, and Hoang Thi Ha Managing Editor: Ooi Kee Beng Editors: William Choong, Lee Poh Onn, Lee Sue-Ann, and Ng Kah Meng Comments are welcome and may be sent to the author(s). |