Be careful about security agencies using AI, govt told

2026-03-22 - 04:40

Earlier this month, home minister Saifuddin Nasution Ismail urged agencies and departments under his ministry to fast-track AI adoption to boost productivity and efficiency. PETALING JAYA: A defence analyst has urged the government to exercise caution as it pushes security agencies to expedite artificial intelligence (AI) adoption in work processes, voicing concerns over its reliability and potential data leakage. Lam Choong Wah of Universiti Malaya said the absence of a Malaysia-owned generative AI forces these security agencies, which often handle important confidential information, to rely on foreign products. Lam Choong Wah. This drives concerns over the security and privacy of data entered into the technology, even for purely administrative purposes, given the lack of domestic safeguards and as the data is stored abroad. “My take is that our security agencies must be extra careful when applying AI. The risk of data leakage cannot be underestimated,” he told FMT. Lam said the reliability and accuracy of AI also remains in doubt, meaning that such tools should be used merely as a reference point. He said there were AI tools in the security sector that could help agencies with their enforcement operations, such as object detectors. However, there are also doubts over their accuracy, said Lam. While AI firms are expected to improve on the reliability and accuracy of their tools in the near future, Lam predicted that they would not be used widely for at least another five years. “It is too premature to claim that AI is already ready to be applied in the security sector.” Earlier this month, home minister Saifuddin Nasution Ismail urged agencies and departments under his ministry to fast-track AI adoption to boost productivity and efficiency. He said generative AI could boost administrative efficiency by analysing large volumes of information and producing draft documents quickly. Heeding the call, the Malaysian Maritime Enforcement Agency (MMEA) said AI would help identify asset strengths and gaps while streamlining operations. Last month, cybersecurity experts urged Putrajaya to ban the use of China’s AI chatbot Deepseek in departments and agencies that handle sensitive data and critical workflows involving national security and sovereignty. They said critical information could inadvertently enter into chatbots whenever prompts are keyed in, with the data possibly stored outside the country and used to train and improve the AI model. Malaysia can’t afford to miss the boat Salawati Mat Basir. Maritime affairs expert Salawati Mat Basir believes AI use in the security sector was practically a must now, saying the coast guards of Southeast Asian nations like Indonesia, Thailand and Philippines are already moving in that direction. Salawati, of Universiti Kebangsaan Malaysia, said AI security tools already exist for intelligence and tactical work, particularly to locate and identify foreign vessels or elements in maritime areas. “AI is being used in wars and conflicts around the world now. Satellites are also used to assist in operations, allowing them to get exact sights of their enemies. “I do believe that AI can boost security operations. If we just stick to what’s conventional, we will not progress,” she said. Acknowledging that Malaysia has fusion centres where satellite imagery is shared between agencies, Salawati said combining these with AI tools could help the MMEA in particular obtain intelligence on threats at sea more quickly. However, she said there was little knowledge about the AI tools that agencies have planned to use or are already using, adding that the information is likely kept discreet in the name of preserving national security. Although a lack of transparency in the past has led to questionable outcomes, Salawati said she was prepared in good faith to give agencies the benefit of doubt.