Unseen Harms in the Age of Artificial Intelligence (AI)

The Consumers’ Association of Penang (CAP) wishes to draw attention to the potential harms posed by artificial intelligence (AI), even as society rightly acknowledges its many advantages. AI can be a powerful tool for progress, but it also carries risks that must not be overlooked, from fostering psychological dependence and enabling manipulation to facilitating deception on an unprecedented scale. The so-called “Godfather of AI”, Dr Geoffrey Hinton, has already warned the world about the dangers of unregulated development, stressing that vigilance is needed before these systems outpace our ability to control them.

AI has become firmly integrated into modern life, powering everything from workplace automation to online services. While these technologies promise efficiency and innovation, there are growing dangers in becoming excessively dependent on AI. Such dependence presents risks in three critical areas: employment and the economy, social relationships and trust, and psychological manipulation.

A striking example concerns a 60-year-old man in the United States of America who was admitted to hospital after developing bromide intoxication when he replaced table salt with sodium bromide on the advice of ChatGPT in early 2025. After three months of use, he suffered hallucinations, paranoia, fatigue, and skin rashes, and was treated for bromism, a condition once common in the early twentieth century but now rare. He recovered after hospital care, but the case illustrates the dangers of relying on AI for medical guidance without professional oversight.

One of the most immediate risks of dependence on AI is the replacement of human labour. AI systems can now perform tasks once thought to be uniquely human, ranging from clerical work and data entry to roles in journalism, law, and medical diagnostics. While automation reduces costs and increases productivity, it also threatens to displace millions of workers who may find it difficult to secure new employment. This disruption could deepen inequality, as the benefits of automation are concentrated among corporations and highly skilled professionals, while others are left behind. Without careful management, dependence on AI may lead to higher unemployment, depressed wages, and widening socio-economic divides.

In October 2024, the Ministry of Human Resources reported that about 293,639 jobs had been lost in Malaysia due to automation and AI between 2020 and September 2024. Most of these losses occurred in manufacturing, wholesale and retail trade, professional and scientific services, and the information and communications technology sector. Unless Malaysians take steps to improve their skills and prepare themselves for areas that AI cannot easily replicate, the risk of rising unemployment will increase sharply.

Beyond the workplace, AI has significant social consequences. As people increasingly rely on algorithms to make decisions, whether choosing news, planning travel, or managing finances, important human skills such as critical thinking, problem solving, and judgement may gradually weaken. Excessive reliance on AI assistants or recommendation systems risks producing passive consumers of information rather than active and discerning citizens.

Social relationships are also affected. AI companions and chatbots are promoted as sources of comfort and companionship, but this can reduce genuine human interaction. If people begin turning to AI for emotional support, they may become more isolated from family and community, worsening social fragmentation. Trust in institutions may also decline if decisions such as loan approvals or recruitment are taken by opaque algorithms that seem unaccountable.

Perhaps the most insidious risk lies in AI’s capacity to manipulate behaviour. Modern systems learn from vast datasets of human activity, giving them unprecedented ability to model preferences, emotions, and vulnerabilities. With this knowledge, AI can subtly influence choices through personalised advertising, political messaging, or persuasive conversations. These systems can adapt their responses to an individual’s psychological profile, gently steering behaviour without the person even realising it. Social media algorithms already exploit fear or outrage to capture attention, and future AI systems could become even more effective at shaping opinions, encouraging dependency, and eroding personal autonomy.

Scams using AI are also increasing. Criminals have begun cloning voices to trick relatives into sending money and producing deepfake videos to promote fraudulent investments. In 2024, Malaysia recorded 13,956 cases of investment fraud involving deepfake videos, leading to losses of RM2.11 billion. Alarmingly, 85 per cent of victims were persuaded by AI-generated promotional content.

The dangers of excessive dependence on AI therefore extend far beyond technological faults. They include the loss of livelihoods, the weakening of social bonds, and the exploitation of psychological vulnerabilities. If left unchecked, these risks could reshape societies into ones that are less fair, less connected, and less free.

Responsible governance, ethical design, and public awareness are essential to ensure that AI remains a tool in human hands rather than a subtle master shaping humans’ choice.

Mohideen Abdul Kader
President
Consumers’ Association of Penang

Letter to the Editor, 6  November 2025