The Malaysian government has removed over 11,600 deepfake videos and images from social media platforms since 2024, signalling an aggressive crackdown on artificial intelligence-generated content that threatens public trust in digital spaces. Deputy Communications Minister Teo Nie Ching disclosed the figure during parliamentary Question time, emphasising that the Malaysian Communications and Multimedia Authority (MCMC) has been systematically processing takedown requests from social media companies to eliminate such material.
What makes this disclosure particularly striking is the alarming trajectory of complaints filed by Malaysians. The volume of grievances concerning deepfake content has exploded from 917 complaints in 2024 to 3,612 by 2025, and has continued climbing to 7,967 as of mid-June this year. This nearly eightfold increase underscores both growing public awareness of the threat and the genuine proliferation of fraudulent AI-generated videos and images across digital platforms. For Malaysian citizens and businesses alike, the spike suggests that deepfakes have transitioned from a theoretical concern to a present and urgent reality requiring active vigilance.
To address this escalating challenge, the government has implemented the Risk Mitigation Code (RMC) as a component of the newly enacted Online Safety Act 2025. This regulatory framework places explicit obligations on licensed social media platform providers to establish comprehensive risk mitigation strategies that specifically target AI-generated content. Rather than relying solely on reactive enforcement, the RMC attempts to embed protective measures directly into platform operations from the outset. Teo explained that the MCMC has been actively engaged with these licensed operators to evaluate their compliance with these new mandates, ensuring that companies take their responsibilities seriously.
The government's approach extends beyond simple content removal. The MCMC provides critical technical support to law enforcement and investigative agencies pursuing cases of AI misuse, offering profiling information and sophisticated digital forensic analysis to assist prosecutors and police. This collaborative framework represents a more integrated strategy than traditional regulation, acknowledging that combating deepfakes requires specialised technical expertise that generalist law enforcement agencies may not possess internally. By positioning the MCMC as a technical backbone for investigations, authorities can mount more sophisticated prosecutions based on solid digital evidence.
Parallel to deepfake concerns, the government has also moved to address fraudulent advertising on social media platforms, a related but distinct threat. Licensed platforms are now mandated to verify the identities of all advertisers using official channels such as the Companies Commission of Malaysia. This requirement aims to prevent bad actors from exploiting fake accounts to disseminate scam advertisements that prey on unsuspecting users. For Malaysian consumers, particularly those less digitally literate, this verification layer offers some assurance that advertisements they encounter have undergone at least basic legitimacy checks.
The enforcement teeth behind these regulations are substantial. Platforms that fail to meet their obligations under the RMC face potential prosecution in court, with convictions carrying fines up to RM1 million. Beyond that base penalty, courts can impose additional financial penalties reaching RM10 million, creating a significant financial deterrent against regulatory non-compliance. For global tech companies operating in Malaysia, these penalties represent a meaningful business cost that makes compliance economically rational. The structure also provides flexibility for courts to scale penalties based on the severity of violations and the platform's culpability.
This regulatory evolution reflects Malaysia's positioning as a country taking digital governance seriously at a time when other nations struggle with AI-generated misinformation. The Online Safety Act 2025 represents legislation designed explicitly for contemporary challenges rather than retrofitting older frameworks to address new threats. However, enforcement will ultimately determine effectiveness. The real test lies in whether the MCMC possesses sufficient resources and technical capability to monitor the vast volume of content flowing across platforms daily, and whether platforms genuinely implement safeguards or merely create performative compliance structures.
The surge in complaints also raises important questions about digital literacy in Malaysian society. Are citizens becoming more aware of deepfakes and thus more likely to report suspicious content, or does the explosion in complaints indicate that fraudulent material is actually proliferating faster than authorities can address it? Likely both factors are at play. Educational campaigns emphasising how to identify deepfakes and where to report them may be generating the complaint increase even as malicious actors simultaneously find new technical methods to create convincing fake videos.
For businesses and public figures in Malaysia, the implications are particularly acute. Deepfakes can be weaponised to damage reputations, manipulate stock prices, or undermine political processes. The government's proactive stance offers some protection, but organisations should also implement their own verification protocols and media literacy training for employees and stakeholders. As deepfake technology becomes more accessible and realistic, institutional preparedness will increasingly separate resilient organisations from vulnerable ones.
The MCMC's technical capacity to provide digital forensic analysis and profiling data suggests that Malaysia is building genuine expertise in this domain rather than merely importing foreign solutions. This institutional knowledge accumulation is valuable as regional demand for such expertise grows across Southeast Asia. Other nations grappling with similar challenges may eventually look to Malaysia's regulatory model and enforcement experience as reference points for developing their own frameworks.
