By Ngo Sheau Shi
After influencer Rajeswary Appahu’s tragic suicide in Malaysia, some are calling for tighter regulations on social media and online content due to recent cyber-bullying incidents.
It may be tempting to introduce new laws, such as a social media licensing law, as a quick fix to address public concerns.
But first, let’s look at Malaysia’s history of handling content moderation and the risks of rushing legislation. It reveals a complex landscape that requires careful navigation.
Double-edged sword of government content moderation
Malaysia’s experience with content moderation has been fraught with problems.
- Sign up for Aliran's free daily email updates or weekly newsletters or both
- Make a one-off donation to Persatuan Aliran Kesedaran Negara, CIMB a/c 8004240948
- Make a pledge or schedule an auto donation to Aliran every month or every quarter
- Become an Aliran member
Historically, regardless of the ruling party, the government has often employed content moderation as a form of moral policing. It is frequently used to silence political dissent while protecting vulnerable individuals.
This pattern will undermine the government’s legitimacy and stability. Valid fears have emerged that legislation could be misused as a tool for political censorship to silence genuine critiques of any abuse of power.
We have witnessed such misuse in the inappropriate employment of draconian laws. These laws include the Sedition Act, the Printing Presses and Publications Act and Section 233 of the Malaysian Communications and Multimedia Act.
Malaysia’s multicultural setting makes it difficult to enforce uniform moral standards through content moderation. Just look at the guidelines for film censorship, announced by Home Minister Saifuddin Nasution Ismail. Such measures make it seem that the “Madani” (clean and compassionate) government is betraying its diverse voters and not keeping its election promise of reform.
This approach could erode public trust and hinder the government’s ability to implement meaningful, long-term changes. It would further alienate the very communities the parties in government had promised to represent and protect.
AI-powered platforms and escalating online toxicity
The quick spread of AI tools on social media makes the online world even more toxic.
The tragic case of Rajeswary Appahu, known as Esha, serves as a stark illustration of the harm that AI-driven social media system can cause. Her story shows how social media algorithms, AI content moderation and real-world consequences can be complicated and risky in Malaysia’s digital world.
Esha, a social media influencer, found herself at the centre of a cyber-bullying storm that led to her untimely death. Her case suggests that AI-driven systems on social media platforms can have unintended consequences. AI algorithms designed to boost engagement might unintentionally amplify controversial or negative content. For instance, if a post criticising Esha gained traction, the platform’s AI might have shown it to more users, thinking it was “engaging” content. This could have snowballed, exposing Esha to an ever-widening circle of criticism and harassment.
AI-driven content moderation systems, while designed to identify and remove harmful content, may struggle with nuanced situations or culturally specific contexts. In Esha’s case, subtle forms of bullying or harassment couched in local slang or cultural references might have slipped through automated filters.
Esha’s case also raises questions about the role of AI in managing viral trends. If negative hashtags or challenges targeting Esha gained momentum, AI systems focused on promoting trending content might have unintentionally contributed to the spread of the cyber-bullying campaign.
Recommendation algorithms, in their effort to show users “relevant” content, might have kept pushing bullying-related content to users who had previously engaged with posts about Esha. This could have created a feedback loop, intensifying the situation.
AI systems are designed with good intentions – to enhance user experience, remove harmful content and promote engaging discussions.
But Esha’s case highlights the potential for unintended consequences in complex, real-world situations.
Her case thus highlights the need for Malaysian policymakers to create AI regulations that take into account the country’s cultural and social dynamics. These could include mandating more transparent AI decision-making processes. These regulations could require platforms to invest in technology to create transparent regulatory compliance.
For instance, social media platforms could be required to provide their content moderation actions on a real-time basis. This would make it easier for a third party to audit their compliance and ensure accountability. This would also help regulatory bodies in Malaysia track and enforce digital regulations in real-time.
The Malaysian government should also think about creating regulatory sandboxes to allow new technologies and business models to be tested in a controlled environment. This would also allow for innovation while ensuring compliance. For instance, local tech start-ups could pilot new AI-driven content moderation tools for hate speech based on the local context and broad-based principles of fairness, ethics, accountability and transparency while still under the supervision of regulatory bodies. These sandboxes can help Malaysia to adapt regulations to the fast-evolving digital landscape.
Esha’s case, while tragic, should serve as a catalyst for positive change in how Malaysia approaches AI regulation and social media governance.
Learning from EU’s AI Act: Need for Asean collaboration
Given the transnational nature of the deployment of AI technologies and the shared challenges faced by Asean member states, a collaborative approach to AI regulation is not just beneficial, it is essential.
This collaborative approach offers several advantages:
- unified bargaining power when negotiating with major tech companies
- shared expertise leading to more robust regulatory strategies
- consistent regional standards to prevent regulatory arbitrage
- increased economic competitiveness by attracting responsible AI investment to the region
I strongly recommend that the Malaysian government, along with other Asean member states, form a multi-stakeholder task force to study and adapt the EU’s AI regulatory framework to the Asean context.
Asean nations can develop a comprehensive and effective approach to regulating AI by learning from global examples and fostering regional collaboration.
The EU’s Artificial Intelligence Act is setting a global standard for AI governance. This groundbreaking legislation establishes obligations for providers, deployers, importers, distributors and product manufacturers of AI systems. It is set to come into effect across all 27 EU member states in the coming month, with most provisions being enforced from 2 August 2026.
The EU’s AI Act offers valuable insights for Asean countries in how it introduces a risk-based approach to AI regulation. The legislation also recognises that, while AI offers immense potential benefits, it also poses major risks if left unchecked.
The cornerstone of the EU’s approach is the categorisation of AI systems based on their potential harm to society. This tiered system comprises four main risk levels: unacceptable, high, limited and minimal risk.
These categories would allow the EU to impose stricter rules and oversight on higher-risk applications while allowing more flexibility for lower-risk ones. This approach ensures proportionate regulation across different AI-use cases. It balances allowing innovation and protecting societal interests.
At the heart of the EU’s AI Act are provisions for “prohibited AI systems”. These are AI applications deemed to pose unacceptable risks to individuals and society at large.
In Malaysia, social media platforms have become an integral part of people’s daily lives. But many of these platforms employ AI systems that deploy manipulative or deceptive techniques. These could distort behaviour and impair informed decision-making, perhaps causing considerable harm.
Esha’s case highlights the pressing need for Malaysia to act swiftly and appropriately.
By leveraging joint resources, Asean can create a unified front to negotiate with big tech companies like Alphabet (Google), Meta (Facebook) and Bytedance (Tiktok). The goal? To establish a comprehensive AI governance framework for the safe deployment of AI technologies.
It is crucial for all stakeholders – platform service providers, regulatory bodies, third-party auditors, legal and technology consultants and civil society representatives – to engage in open dialogue about how to create safer online spaces. Each stakeholder has a specific role to play in contributing to checks and balances within the system.
Dr Ngo Sheau Shi, an Aliran member, teaches at the School of Communication in Universiti Sains Malaysia in Penang.
AGENDA RAKYAT - Lima perkara utama
- Tegakkan maruah serta kualiti kehidupan rakyat
- Galakkan pembangunan saksama, lestari serta tangani krisis alam sekitar
- Raikan kerencaman dan keterangkuman
- Selamatkan demokrasi dan angkatkan keluhuran undang-undang
- Lawan rasuah dan kronisme
