Southeast Asia's AI edge lies in reading diverse risks and collective influence in global governance

The same artificial intelligence (AI) that helps governments automate services, analyse data, and detect cyber threats can be weaponised against the very systems it's meant to protect. 

For governments in Southeast Asia, the challenge lies in harnessing AI’s capabilities while guarding against ever-evolving threats.  

For the US-based Institute for AI Policing Strategy (IAPS)’s senior researcher, Jam Kraprayoon, he believes that ASEAN’s role is not to compete head-on with AI powers, but build institutional capability through its diverse challenges.

“Cybersecurity risks from AI are shared globally,” said  Kraprayoon, highlighting that “ASEAN may face greater exposure because the general cybersecurity hygiene baseline is lower in many of these countries.”  

While global leaders may shape the frontier of AI development, he believes Southeast Asia has the advantage in understanding how AI risks impact the ground. 

AI safety is becoming a challenge of state capacity 

Global AI governance is largely shaped by the US, Europe and China. 

The US ecosystem is driven by frontier AI companies, while Europe focuses on regulation through the EU AI Act. China, on the other hand, adopts a more state-directed approach.  

Hence, what works for the countries might not be directly adoptable in ASEAN, especially when each member state differs in economic development, digital maturity, cybersecurity readiness and resources availability. 

“Countries should look at practices that fit their context and balance them with what is possible given their positioning,” Kraprayoon notes. 

ASEAN’s strength in governing AI in diverse environments 

The region’s experience with cross-border scam operations, multilingual societies, and uneven digital ecosystems can provide practical lessons for developing safety measures that work in the real world. 

“If ASEAN countries can create a common voice, they are more likely to influence things in a positive direction,” Kraprayoon says. 

The cross-border risks have called for synergy across the member states.   

Singapore's AI Safety Institute is building technical evaluation capability by collaborating with global researchers to assess advanced AI systems. 

Malaysia, during its ASEAN Chairmanship, spearheaded the establishment of the ASEAN AI Safety Network, followed by its own AI Safety Institute. 

Thailand has expanded its AI Governance Centre into the AI Governance Practice Center (AIGPC), focusing on capacity building, risk testing and rolling out AI governance playbooks. 

This reflects one of ASEAN's greatest strengths, as Southeast Asia becomes a testing ground for solving AI safety challenges across different languages, cultures and operating environments. 

“A lot of these concerns will be much weaker if spoken through by one country.  

“But if ASEAN can bargain collectively, given the large shared market, it is more likely that they will be able to influence things in a positive direction,” Kraprayoon shares.

The shift from AI adoption to agentic governance 

The stakes become higher as AI moves from assisting humans to acting with greater autonomy. 

For public institutions, the shift from AI assistants to AI agents could pose cyber risks as the latter can now access more information, execute tasks across systems, and exploit the cyber vulnerabilities.  

This prompts the organisation to question who remains accountable when AI systems act with greater autonomy. 

For that reason, Korea and Singapore AI Safety Institutes, alongside the Singapore AI Safety Hub are now working on AgentID project, a tool to better identify and manage AI agents. 

Greater autonomy could allow governments to respond to threats at machine speed. But it also leaves less time for human intervention when something goes wrong. 

“The tempo of operations dictates whether humans can remain in the loop,” Kraprayoon highlights. 

As AI becomes more embedded in government operations, safety will increasingly become a national resilience issue involving cybersecurity, critical infrastructure, and public trust.  

Human capability remains the final safeguard 

For civil servants, AI literacy cannot stop at knowing how to use a chatbot.  

It requires understanding when AI should be trusted, when it should be challenged, and how its output influences public decisions. 

“Civil servants must develop their own understanding of frontier AI, particularly where it intersects with cybersecurity, geopolitics and national resilience,” he adds.  

Although ASEAN does not need to replicate US or China’s scale in frontier AI development, it needs sufficient technical capability to evaluate and govern advanced systems. 

As Kraprayoon shares, “The future of AI governance will not be determined solely by those building the most powerful models. It will also be shaped by governments capable of understanding, questioning and governing them.”