PETALING JAYA: Malaysian Home Minister Saifuddin Nasution Ismail has called on security agencies to accelerate the adoption of artificial intelligence (AI) to enhance productivity and efficiency. However, experts are urging caution as the government pushes for swift implementation, highlighting concerns over data security and the reliability of AI systems.
Defence analyst Lam Choong Wah from Universiti Malaya warned that the reliance on foreign AI products, due to the absence of a domestically developed generative AI, poses significant risks, particularly for agencies that manage sensitive information. He emphasized the potential for data leakage and the importance of exercising caution when integrating AI into security protocols.
“The risk of data leakage cannot be underestimated,” Lam stated, noting that the reliability and accuracy of AI technologies remain in doubt. He suggested that AI tools should be viewed as references rather than definitive solutions, particularly in security contexts where precision is paramount.
Despite acknowledging the existence of AI tools that could assist in enforcement operations—such as object detectors—Lam expressed skepticism regarding their effectiveness. He forecasted that widespread adoption of AI in the security sector might take at least five years, indicating that current technologies are not yet adequately prepared for such critical applications.
Earlier this month, Saifuddin urged agencies under his purview to leverage generative AI for tasks like analyzing large datasets and drafting documents more swiftly. His call was echoed by the Malaysian Maritime Enforcement Agency (MMEA), which suggested that AI could help identify operational strengths and weaknesses, thereby streamlining processes.
However, the push for AI adoption has raised alarms among cybersecurity experts. Last month, they recommended that the government prohibit the use of China’s AI chatbot Deepseek in departments handling sensitive or critical workflows related to national security, citing the risk of inadvertently exposing confidential information stored abroad.
Maritime affairs expert Salawati Mat Basir, from Universiti Kebangsaan Malaysia, disagrees with a purely cautious approach. She noted that neighboring countries like Indonesia, Thailand, and the Philippines are already integrating AI into their coast guard operations, underscoring the necessity for Malaysia to keep pace or risk falling behind.
“AI is being used in wars and conflicts around the world now,” Salawati stated, highlighting the use of satellites for tactical advantages. She believes that AI has the potential to greatly improve security operations, and that a rigid adherence to conventional methods would hinder progress.
Salawati acknowledged the existence of fusion centers in Malaysia that facilitate the sharing of satellite imagery among agencies. She suggested that combining these resources with AI tools could enable the MMEA to gather intelligence on maritime threats more swiftly and effectively.
Despite her optimism, she pointed out a lack of transparency surrounding the AI tools currently in use or planned for deployment by security agencies. Salawati remarked that while past confidentiality has led to questionable outcomes, she is willing to give agencies the benefit of the doubt as they navigate these advancements.
The adoption of AI in Malaysia’s security framework raises complex questions about the balance between innovation and safeguarding sensitive data. As the government pushes forward, it must navigate these challenges carefully to ensure the security and privacy of its citizens while remaining competitive in a rapidly evolving technological landscape.
See also
Geoffrey Hinton Warns of AI Misuse and Existential Risks Amid Rapid Advancements
Employers Must Address AI Cybersecurity Risks to Protect Retirement Assets, DOL Warns
Bosnia and Herzegovina Tops List of Countries Most Vulnerable to AI Cyberattacks
Accenture and Microsoft Launch AI-Powered MxDR for Enhanced Cybersecurity Resilience
AI Enhances Cybersecurity, Automating Threat Detection but Requiring Human Expertise
