AI doesn’t negotiate: Will Malaysia rule it or obey it? — Ahmad Nizam Che Kasim

September 16 - The United Nations High Commissioner for Human Rights, Volker Türk, recently issued a stark warning about the dangers of unchecked artificial intelligence, stating that AI-related harms are "already occurring, front and centre". Türk's concerns include discrimination, algorithmic exclusion, mass surveillance, and AI-driven disinformation, which are already straining democratic institutions.

As autonomous AI systems become increasingly capable of pursuing objectives beyond their original design, Türk warns of "a step change towards greater existential risks to every aspect of our lives". This assessment comes from the Independent International Scientific Panel on Artificial Intelligence, and underscores the need for swift AI governance to keep pace with technological development.

Malaysia, currently aiming to become an "AI Nation" by 2030, finds this warning particularly relevant as it invests heavily in AI infrastructure and governance through initiatives such as its National AI Action Plan 2026 to 2030 and a RM2 billion sovereign AI cloud. While Malaysia's actions are commendable, Türk's letter emphasizes that the speed of AI adoption must not outpace the pace of AI governance.

To address this, Malaysia must implement enforceable, independent governance measures, such as independent verification of AI systems, mandatory reporting of serious incidents, and enforceable safety plans backed by liability regimes. These measures must also ensure that responsibility for AI actions always rests with identifiable humans, rather than the machines themselves.

While Malaysia's existing plans show some progress in these areas, they lack the necessary teeth, such as real audit access, penalties, and independence from ministries promoting AI adoption. Furthermore, the reality is that the frontier of AI development is being largely shaped by the United States and China, not by Malaysia. This highlights the importance of multilateral engagement, particularly through international forums like the Global Dialogue on AI Governance, for Malaysia to advocate for human rights protections in AI decision-making.

Türk concludes his letter with the important reminder that no one should have to sacrifice their human rights for the sake of technological progress. Malaysia can indeed pursue its AI ambitions while respecting this warning, but only if governance is treated as a critical precondition for growth, not an afterthought.