Deepfakes, cyberattacks among five AI threats identified in Malaysia

KUALA LUMPUR: Deepfakes, cyberattacks and data privacy breaches are among five major threats posed by artificial intelligence (AI) identified in Malaysia, raising concerns over the misuse of the rapidly evolving technology.
The other threats are algorithmic bias and the misuse of generative AI, according to findings from a series of AI safety and security roundtable discussions involving local and international experts.
Organised by AI Malaysia as part of the development of the National AI Action Plan 2026–2030, the discussions highlighted the growing need for safeguards to ensure AI systems remain under human control.
The National Artificial Intelligence Office (NAIO) said concerns over humans gradually losing control of AI should be examined from the perspectives of safety, governance, accountability and oversight.
It said these concerns had become increasingly relevant as AI systems grew more sophisticated and complex.
NAIO said the National AI Action Plan 2026–2030 recognised that regulatory oversight alone would not be sufficient to address emerging risks.
As such, the plan proposed developing dedicated capabilities to assess, monitor and respond to AI-related threats, including through a proposed AI Trust Function.
"This approach is aligned with the human-centred AI principles underpinning AI Nation 2030.
"The action plan stipulates that AI adoption must strengthen human well-being, respect societal values and integrity, and be implemented through responsible governance," it said.
NAIO said Malaysia's priority was not based on the assumption that AI would eventually take complete control over humans.
"Rather, the focus is on ensuring that humans and institutions retain clear control, oversight and accountability over how AI is developed and used," it said.
On whether the government would introduce restrictions on certain AI applications, NAIO said Malaysia had yet to enact legislation specifically regulating the technology.
It said there was also no definitive list of prohibited AI applications, as the proposed AI Governance Bill was still being developed.
However, the existing AI Governance and Ethics Guidelines (AIGE) outlined principles covering safety, privacy, security, transparency, accountability and human well-being.
NAIO said the proposed legislation would not replace existing laws governing criminal activities involving AI.
"Any use of AI involving activities that are already unlawful will remain subject to the relevant laws and enforcement frameworks.
"For example, the misuse of AI to commit cybercrime or offences involving communications and content will continue to be governed by existing laws," it said.
These include legislation relating to cybercrime and communications and multimedia, depending on the nature of the offence.
On calls for a minimum age of 16 for AI use, similar to restrictions introduced for social media, NAIO said such a move should not be considered without examining the differences between the two technologies.
It said AI applications served different purposes and carried varying levels of risk, making a blanket age restriction potentially unsuitable.
"Any future age-related restrictions should consider the type of AI application, its risks and the circumstances in which it was used."
NAIO said the government's current priority under the National AI Action Plan 2026–2030 was to strengthen AI literacy among primary and secondary school students.
This included educating students about AI safety, ethics and the responsible use of the technology.
KioskNews shows a cleaned-up reading view extracted from the publisher’s page — the original always lives on their site, not ours.