Malaysia's fight against artificial intelligence-generated deception has entered a more aggressive phase, with the government confirming that over 11,600 deepfake videos and images have been taken down from social media platforms since the start of 2024. Deputy Communications Minister Teo Nie Ching disclosed these figures during Parliamentary question time, underlining the severity of a phenomenon that has accelerated dramatically across Southeast Asia's largest digital economy.
The trajectory of complaints filed with Malaysian authorities paints a stark picture of how rapidly deepfake technology has been weaponised by bad actors seeking to deceive the public. Starting from 917 complaints in 2024, the figure ballooned to 3,612 cases in 2025 before reaching 7,967 reports as of mid-June this year—a multiplication that suggests deepfake creation and distribution have become increasingly normalised as perpetrators test the boundaries of enforcement. This eightfold surge reflects both growing public awareness of the threat and a corresponding uptick in actual malicious activity, with the technology becoming more accessible and sophisticated.
These removals have been achieved through a coordinated mechanism whereby the Malaysian Communications and Multimedia Authority submits takedown requests directly to major social media platforms, leveraging the platforms' obligation to respond swiftly to government directives. The efficiency of this process depends on platform cooperation and technical capacity to identify and verify deepfake content, a challenge that grows exponentially as AI generation tools become more realistic. The fact that platforms are responding to these requests underscores the leverage that regulatory frameworks can exercise, even in an environment where global tech companies jealously guard their operational autonomy.
The legal architecture underpinning this enforcement effort is the Online Safety Act 2025 and its associated Risk Mitigation Code, which imposes explicit obligations on licensed social media platforms to implement safeguards against AI-generated content abuse. This represents a philosophical shift in Malaysian regulatory thinking—from reactive punishment of harm to proactive prevention through platform-level controls. Rather than waiting for deepfakes to circulate and cause damage, the framework requires platforms to build detection and mitigation systems into their operations, making them gatekeepers of artificial content integrity.
Under these new provisions, licensed platforms must establish risk mitigation measures specifically targeting artificially generated materials, with the Malaysian Communications and Multimedia Authority conducting ongoing assessments to verify compliance. This supervisory role requires technical expertise that government agencies are developing in real time, as the AI arms race between creators and detectors accelerates globally. The MCMC's engagement with platforms represents perhaps the first systematic attempt in Southeast Asia to mainstream AI governance directly into platform operations, creating a model that other regional governments may adapt.
Beyond takedowns, the Malaysian authorities have adopted a multi-pronged approach that combines technical investigation with platform accountability. The MCMC provides forensic support to law enforcement agencies, offering digital analysis and profiling information that can help trace the origins of deepfakes and identify perpetrators. This technical assistance bridges the gap between platforms' detection capabilities and investigators' evidentiary needs, ensuring that removals translate into potential prosecutions rather than mere content erasure. For a country still developing its cybercrime investigation infrastructure, this represents meaningful capability-building.
A related dimension of platform abuse that Teo addressed involves fraudulent advertising, where scammers exploit social media's commercial functions to deceive consumers. The regulatory response mandates identity verification for advertisers, requiring platforms to cross-reference advertiser information with official registries such as the Companies Commission of Malaysia. By preventing the creation of anonymous or false commercial accounts, this measure cuts off a common pathway for scammers to reach potential victims at scale. The requirement signals that platform responsibility extends beyond content moderation to the economic integrity of the advertising ecosystem itself.
The penalties for non-compliance are substantial enough to concentrate platform attention. Licensed social media operators that breach their Risk Mitigation Code obligations face potential court prosecution, with fines reaching RM1 million for the violation itself, plus additional financial penalties of up to RM10 million. These figures are calibrated to be meaningful for commercial enterprises but not so catastrophic as to drive platforms from the market—a careful balance that reflects Malaysia's desire to maintain a functional digital economy while imposing genuine consequences for negligence or bad faith.
The escalation of enforcement reflects broader anxieties about deepfakes' political and social consequences across Southeast Asia. In a region where election cycles continue and social cohesion remains contested, synthetic media depicting political figures in compromising situations or inciting communal violence presents existential risks to institutional stability. Malaysia's regulatory approach is therefore not merely about consumer protection; it is fundamentally about defending the informational commons from weaponised artificial content that could undermine public trust in institutions and between communities.
For Malaysian consumers and businesses, these developments carry practical implications. Users encountering suspicious video or audio content have clearer pathways for reporting through official channels, while companies face greater pressure to verify the authenticity of claims made in marketing materials. The regulatory framework also creates opportunities for Malaysian technology companies to develop detection and authentication tools that comply with local standards, potentially seeding a domestic AI governance industry.
Regionally, Malaysia's experience offers a case study in how middle-income democracies with functional digital regulation can adapt quickly to emerging technological threats. The speed with which the Online Safety Act framework was operationalised and the explicit focus on AI governance contrast with slower international efforts to establish global deepfake standards. As other Southeast Asian nations grapple with similar challenges, Malaysia's regulatory blueprint—combining platform obligations, government technical capacity, and graduated penalties—may serve as a template.
Looking forward, the sustainability of Malaysia's deepfake response depends on whether platforms continue investing in detection as AI generation tools grow more sophisticated, and whether government agencies can retain technical talent in a competitive global market for AI expertise. The surge in complaints suggests that public awareness is improving, but so is the ease of creating convincing fakes. The coming years will test whether Malaysia's regulatory framework can keep pace with technological change, or whether constant amendments will be necessary to maintain effectiveness in an environment where deepfake capability doubles yearly.
