MALAYSIA MUST STRENGTHEN AI POLICIES, REGULATORY FRAMEWORK TO CURB DEEPFAKE SCAMS

By Mohamad Syazwan Mustafa

JOHOR BAHRU, March 29 (Bernama) -- Malaysia must urgently strengthen its artificial intelligence (AI) policies and regulatory framework, especially for high-risk technologies, to protect digital security and combat the growing misuse of AI, including deepfake scams.

Institute for Data Innovation and Artificial Intelligence (IDEA-AI) chief executive officer and director Prof Dr Mohd Saberi Mohamad, said this is crucial as AI technology becomes increasingly sophisticated, capable of mimicking voices and identities to deceive victims, driving a significant rise in online fraud.

bootstrap slideshow

He noted that over 35,000 online fraud cases were reported in 2024, resulting in losses of approximately RM1.58 billion, while the first three months of 2025 saw 12,000 cases amounting to RM573.7 million.

“While not all cases involve deepfakes, AI is increasingly being used by fraud syndicates to impersonate voices or identities to mislead victims.

“Globally, cybersecurity reports indicate that deepfake-based fraud has increased more than tenfold between 2022 and 2023,” he told Bernama.

Last November, the National Artificial Intelligence Office (NAIO) reported that it is developing a regulatory framework for AI, covering risk management, harm mitigation, incident reporting and ethical principles to underpin forthcoming legislation.

The Ministry of Digital is also finalising the AI Governance Bill, designed to tackle AI risks comprehensively and ensure a safe, responsible ecosystem.

Mohd Saberi, based in Melbourne, Australia, and Kuala Lumpur, noted that existing laws, including the Communications and Multimedia Act 1998 and the Penal Code, were drafted before the rise of deepfake technology and do not specifically address AI.

“New situations are often hard to interpret. The lack of dedicated AI legislation, difficulties in identifying perpetrators, cross-border jurisdiction issues, and insufficient evidence standards pose major enforcement challenges,” he said.

He recommended a risk-based regulatory approach: high-risk AI applications, such as those in safety, finance, biometric identification, and automated decision-making, should face strict controls, while low-risk uses are given space to innovate.

“Digital literacy among citizens must be strengthened. AI policy committees should balance senior and junior members. An AI Act should be enacted to define developers’ and users’ responsibilities, preventing the misuse of technologies like deepfakes,” he said.

Meanwhile, Dr Nooraziah Ahmad, associate fellow and senior lecturer at Universiti Malaysia Kelantan’s Institute for Artificial Intelligence and Big Data, warned that the widespread circulation of AI-generated content and misinformation could spark panic and exacerbate racial, religious, or political tensions.

“If public figures, government agencies, or financial institutions can be impersonated via deepfakes, public trust in national institutions will erode. Laws should target fraud, extortion and political manipulation, while encouraging AI use for education, business and innovation,” she said.

She added that a clear, consistent governance framework is needed to give industry regulatory certainty, attract investment and ensure Malaysia remains a safe, innovation-friendly environment.

“High-risk AI applications may require licensing, balancing public safety with innovation. This risk-based approach aligns with international frameworks like the EU AI Act to control harmful AI use,” she said.

-- BERNAMA