The Malaysian government has intensified its crackdown on artificial intelligence-generated deception, removing over 11,600 deepfake videos and images from social media platforms following formal takedown requests submitted by the Malaysian Communications and Multimedia Commission. Deputy Communications Minister Teo Nie Ching disclosed the scale of enforcement action during parliamentary proceedings, underscoring a nationwide pivot towards aggressive regulation of AI-enabled content manipulation.
The trajectory of complaints filed against deepfake material reveals the accelerating nature of the threat facing Malaysian society. Starting from 917 documented cases in 2024, the numbers escalated dramatically to 3,612 complaints in 2025 before climbing further to 7,967 cases by mid-June this year. This eightfold increase in less than 18 months demonstrates both a growing public awareness of deepfake dangers and a corresponding rise in actual instances of malicious AI-generated content circulating on digital platforms. The trend suggests that without ongoing intervention, deepfakes could become an entrenched feature of Malaysia's information landscape, undermining trust in media, political discourse, and commercial transactions.
The regulatory framework underpinning these takedown operations rests on Malaysia's relatively new Online Safety Act 2025, which introduced the Risk Mitigation Code as a binding set of standards for licensed social media operators. Rather than relying solely on reactive removal of problematic content, this legislative approach mandates that platforms implement proactive safeguards specifically targeting AI-generated material. The shift represents a departure from earlier, more permissive regulatory models and signals Kuala Lumpur's determination to position Malaysia as a leader in responsible AI governance within the Southeast Asian region.
Implementation of the RMC has involved systematic engagement between the MCMC and major social media providers operating within Malaysian jurisdiction. Government officials have been tasked with auditing how effectively these platforms are meeting their obligations to detect, flag, and remove AI-generated content before it achieves widespread circulation. This oversight mechanism addresses a fundamental weakness in self-regulation: platforms' historical reluctance to invest heavily in AI detection when the financial incentives favour engagement metrics over safety. By making compliance a condition of licensure in Malaysia, the government has converted platform security from a corporate discretion into a legal requirement.
Beyond removal operations, the MCMC has developed technical capabilities to support law enforcement investigations into deepfake-related crimes. By conducting digital forensic analysis, profiling suspected creators, and providing evidentiary support to police and regulatory bodies, the commission functions as both a detection agency and an investigative partner. This dual role is particularly important in cases where deepfakes serve as instruments of fraud, defamation, political manipulation, or harassment. The technical barriers to prosecution have historically been high—establishing chains of custody for digital evidence, identifying perpetrators concealed behind proxy servers and VPNs, and attributing creation tools to specific individuals all present formidable challenges that generalist law enforcement agencies struggle to overcome.
The deepfake problem intersects with broader concerns about election integrity and political stability. Malicious actors can weaponise convincing AI-generated video to damage candidates, sow discord between communities, or delegitimise democratic processes. Several democracies have experienced targeted deepfake campaigns preceding elections, creating scenarios where voters cannot reliably distinguish authentic from fabricated content. Malaysia's experience with political polarisation and religious sensitivity makes the jurisdiction particularly vulnerable to deepfakes designed to incite communal conflict or undermine confidence in elections. The government's enforcement push thus carries implications extending well beyond ordinary consumer fraud into the realm of national security.
Parallel to deepfake enforcement, the government has targeted fraudulent advertising schemes exploiting platform visibility. Licensed platforms are now required to implement advertiser verification systems, cross-referencing submissions against databases maintained by the Companies Commission of Malaysia and other official registries. This requirement aims to prevent criminals from establishing fake business accounts and using platform advertising tools to promote scams targeting Malaysian consumers. The measure acknowledges that platform architecture itself—designed to maximise monetisation through advertising volume—creates inherent vulnerabilities that bad actors can easily exploit unless substantial friction is introduced into the verification process.
Enforcement teeth have been sharpened through penalties embedded in the Online Safety Act framework. Platforms found to have failed their RMC obligations face potential prosecution and financial sanctions reaching RM1 million in base fines, with additional penalties accumulating to RM10 million in aggravated cases. These sums represent meaningful exposure for platform operators, sufficient to capture board-level attention and justify substantial compliance investments. The tiered penalty structure also creates incentive gradation: minor breaches attract proportionate penalties, while systematic failures to implement required safeguards trigger the maximum financial consequences. This calibration attempts to balance the imperative to deter non-compliance against the risk of driving major platforms entirely out of the Malaysian market.
The legislative approach reflects lessons learned globally regarding the inadequacy of purely corporate self-regulation in managing AI-generated content harms. The European Union's Digital Services Act, similar regulatory moves in Singapore and Australia, and proposals under consideration in other democracies all converge on mandating platform accountability through legal requirements rather than relying on industry goodwill. Malaysia's adoption of this model positions the country alongside more developed economies in regulatory maturity, potentially enhancing its standing as a responsible digital marketplace attractive to international advertisers and platforms seeking jurisdictions with clear legal frameworks.
Looking forward, the effectiveness of Malaysia's deepfake enforcement framework will depend on sustained investment in technical capacity, consistent application of penalties, and ongoing evolution of detection methodologies to keep pace with advancing AI capabilities. Generative AI models continue to improve rapidly, and deepfake creation tools are becoming increasingly accessible to non-specialist users. The arms race between deepfake generation and detection technologies may well determine whether Malaysia's current enforcement surge succeeds in deterring misuse or merely impedes the most careless perpetrators while sophisticated actors continue undetected. Coordinating efforts with regional partners through ASEAN and international bodies could amplify enforcement effectiveness while spreading the technical burden across multiple jurisdictions confronting identical threats.
