.png)
SHAH ALAM – Government agencies should not fully rely on artificial intelligence (AI) for critical public safety campaigns as errors could arise without human oversight, says communications expert Datuk Dr Khairuddin Othman.
The former Faculty of Communication dean at HELP University said AI should only serve as a supporting tool rather than replace human judgement, especially when producing materials that provide important information to the public.
He said the complete dependence on automated systems for safety-related communication could result in serious consequences if errors go undetected.
"The total reliance on automation for critical safety messaging introduces severe failure instances that can undermine public trust and worse, cause real-world harm," he said.
Khairuddin said one of the major concerns surrounding AI-generated content was its tendency to produce inaccurate information, including fabricated details or incorrect instructions.
"There is already substantive knowledge that AI models generate factual errors, fabricate data, or invent safety protocols. In a crisis, providing an incorrect emergency phone number or an ineffective rescue effort can be fatal," he said.
He said public safety campaigns also depend heavily on credibility and audiences could lose confidence when they encounter unnatural visuals, audio or other errors in AI-generated materials.
"Public safety relies heavily on institutional authority and community trust. Malaysian audiences are not accustomed to fully AI-generated synthetic pictures and voices.
"They often reject the message once they notice visual glitches or unnatural audio, leading them to ignore the actual safety warning," he said.
Khairuddin said organisations should adopt a "human-in-the-loop" approach, where human experts remain responsible for reviewing and approving AI-generated materials before they are released.
Every piece of text, image and video produced using AI, he said, should undergo proper evaluation by subject matter experts to ensure factual accuracy, cultural and religious sensitivity, as well as clarity of the message.
He added that humans should also remain in control of campaign delivery, including deciding the target audience and scheduling of advertisements, rather than allowing AI systems to fully automate the process.
His remarks come amid increasing scrutiny over the use of AI in public communication, particularly after an alleged AI-generated cinema advertisement on drowning rescue protocols drew criticism over multiple alleged factual and visual inaccuracies.
The issue sparked wider debate over whether public-facing materials, especially those involving safety information, should undergo stricter human review before being released.
The concerns surrounding AI-generated content are not new, with previous incidents in Malaysia highlighting the risks of relying on automated systems without proper verification.
In 2024, an AI-generated digital billboard in Desa Sri Hartamas, Kuala Lumpur, mistakenly depicted the Petronas Twin Towers with three towers instead of two.
In 2025, the Education Ministry also faced criticism after publishing an AI-generated cover image for its Sijil Pelajaran Malaysia (SPM) results analysis report that showed a Jalur Gemilang with two stars instead of one and eight stripes instead of the official 14.
These incidents have fuelled wider discussions on the balance between technological efficiency and human responsibility.
While AI can help organisations produce content more quickly, human oversight remains important to ensure information shared with the public is accurate, reliable and appropriate.
