Artificial intelligence (AI) is no longer merely a technology issue. It is fast becoming the new terrain of great-power rivalry.

The latest Anthropic paper on global AI leadership in 2028 makes this clear: the contest between the United States and China is no longer only about trade, tariffs, chips or military alliances. It is about who will shape the intelligence infrastructure of the world. For the lack of a better word, AI.

Yet AI, whether it is built in US or China, or for that matter,  the European Union (EU), will carry all their confirmation bias prejudiced. Each of these three entities will build their AI programmes based on their worldviews.

For example, a simple prompt of ChatGPT on Asean, which is built in the US, will describe Asean as a region based on a consensus-driven model that slows all things down.

To the degree China or EU’s AI have their own prejudice against Asean, that form of negative optic will also emerge.

In due course, Asean citizens that consume all three versions of AI will not understand the importance of the Zone of Peace, Freedom and Neutrality (Zopfan) which are slow to evolve yet strategically powerful and vital to keep the region safe from intellectual or epistemic whitewash of its own intellectual tradition.

For Asean, and especially Malaysia, this is a warning. If Southeast Asia is careless, it will not merely use AI.

It will be used by AI systems, standards, cloud infrastructures and security doctrines designed elsewhere.

Anthropic frames the future as two possible scenarios. It is vital to understand this is still an understatement of the problem of AI of US and China since the problem posed by Anthropic, which works with the Pentagon, is still a paper written from the rendition of what is democratic (US) and what is authoritarian (China).

Yet the true reality of AI of these two polities is more nuanced than a binary choice. AI users must build up the residence and digital sophistication to look through and beyond their systemic bias.

In the first, the United States and its democratic allies maintain a commanding lead in AI through advanced chips, export controls, trusted infrastructure and global adoption. Malaysia and Asean should understand their promise and perils.

In the second, China catches up through computing loopholes, cheaper models, rapid state adoption and large-scale model distillation. Yet this system is not perfect either.

Again, Malaysians and citizens of Asean should learn to go beyond the worldview of any one entity.

In either case, if Asean is not, jointly or separately, the main author of the story, the bloc and its member states risk the fate of “digital colonialism”.

This is precisely what Malaysia must avoid. The current government has come up with Ilmu AI to defray the effects of Western or Chinese AI.

The point is not that Asean should reject AI. That would be impossible and self-defeating.

AI will transform education, healthcare, logistics, finance, agriculture, cybersecurity and public administration.

Malaysia’s digital economy, data centres, semiconductor ecosystem and universities can all benefit.

But the region must not enter AI passively, as it once entered parts of the global supply chain: useful, cheap, connected, but strategically dependent.

The new AI rivalry is built on four fronts: intelligence, adoption, global distribution, and resilience.

For what it is worth, Anthropic rightly notes that the most advanced AI models will shape not only markets but also political norms and security systems.

This means that countries that do not build their own AI governance capacity will import someone else’s values, biases and strategic priorities.

For Malaysia, the challenge is even sharper. Malaysia is a trading state, a semiconductor node, an energy producer, a data centre destination and Asean’s natural diplomatic convenor. It cannot afford to choose technological naivety.

Nor should it be forced into a crude choice between Washington and Beijing.

Instead, Malaysia must lead Asean toward an AI non-alignment doctrine.

Such a doctrine should have five pillars.

First, Asean must build sovereign AI literacy across government, universities, and industry.

Second, Malaysia must strengthen the awareness of the danger of “epistemic dominance”. All systems must be based on what is free and open as enshrined in Free and Open Indo Pacific Strategy (FOIP) and Asean Outlook on Indo Pacific (AOIP).

In other words, Malaysians and citizens of Asean who increasingly use foreign AI would lead them to collective indoctrination.

Third, Asean must insist that foreign AI infrastructure operating in the region is transparent, auditable and compatible with local laws.

Fourth, Malaysia should build trusted regional datasets that do not obscure the achievements and style of the Asean diplomatic achievements to safeguard the digital sovereignty of the bloc.

As and when there are any shortcomings, Malaysia, as a hub of data centres in Asean must have think tanks such as the likes of the Institute of International and Asean Studies (IINTAS) working hand in hand with those in University of Tokyo, even Seoul National University, or, National University of Singapore, to warn the rest of the think tanks of the emerging risks and the concurrent responses that are needed.

Fifth, Asean must create an AI diplomacy platform under Asean Plus Three, the East Asia Summit and the Asean Digital Ministers’ Meeting.

The point is not to resist America or China. The point is to prevent Asean from becoming technologically colonised by either.

Anthropic’s paper is written from an American strategic perspective. That is understandable. But Asean must read it with Asean eyes. As does Malaysia with our national interest in mind.

If the US sees AI leadership as a democratic-security imperative, and China sees AI as a national rejuvenation imperative, Asean must see AI as a sovereignty imperative.

Malaysia’s task is therefore urgent. It must not wait until 2028 to discover that the rules have already been written.

AI governance, computer access, cybersecurity, education reform and public-sector adoption must be treated as national strategy, not technical housekeeping.

The future AI order will not be neutral. It will reflect the power of those who build it.

The views expressed are those of the writer and do not necessarily reflect those of FMT.