Is AI causing Singapore to lose its edge?

On 13 June 2026, at 5:21pm Eastern Time, Anthropic received a directive from the US government. By that evening, Fable 5 and Mythos 5 — the most capable AI models publicly available anywhere in the world — were taken offline for every user outside the United States. No notice. No appeal process. No timeline for restoration. The directive cited a national security concern: a potential method of bypassing the model's safeguards.

Anthropic complied, as it was legally required to do. It also disagreed — publicly and clearly — stating that the vulnerability disclosed was narrow, non-universal, and already present in other models freely available on the market. The models went dark anyway.

This was not a technical failure. It was a policy decision made in Washington that instantly affected hundreds of millions of people and organisations across Europe, Asia, and the rest of the world who had built workflows, products, and capabilities on top of those models. They had no vote. No recourse. No warning.

01What actually happened — and what it means

To be precise about what occurred: the US government issued an export control directive prohibiting access to Fable 5 and Mythos 5 by any foreign national — whether inside or outside the United States. Because Anthropic could not selectively disable the models for foreign nationals without disabling them for everyone, the models were suspended globally.

Anthropic's disagreement is notable but beside the point for anyone outside the US. The company complied. The models went down. And the lesson was not about jailbreaks or cybersecurity — it was about concentration of power.

When the world's most important AI infrastructure is owned, operated, and regulated by a single country, that country's domestic politics become everyone else's operational risk. A national security concern in Washington becomes a service disruption in Singapore. A policy disagreement between an AI company and an administration becomes a capability gap for businesses in London, Tokyo, and Seoul that had nothing to do with it.

02The rest of the world is paying attention

At the G7 summit in Evian-les-Bains on 17 June 2026, France's President Emmanuel Macron made what observers described as a forceful plea for the United States not to keep cutting-edge AI to itself. He characterised the US government's intervention with Anthropic as a "strictly nationalist" reaction — and pushed for a "trusted partners" framework that would give allied nations guaranteed access to frontier models, regardless of domestic US policy decisions.

The European Commission had moved first. On 3 June — ten days before the Fable 5 shutdown — it unveiled a comprehensive tech sovereignty package: Chips Act 2.0, a Cloud and AI Development Act, and an open-source strategy. The stated goal was explicit and unambiguous.

The EU is committing hundreds of billions of euros — an estimated €100 billion for cloud and AI alone — to build the infrastructure needed to run European AI on European compute, governed by European law. They are not asking permission. They are building.

03Singapore's exposure

We need to be honest about something that tends to get softened in polite conversations about AI policy: Singapore is heavily exposed to this dynamic, and we are not moving fast enough.

Singapore has no natural resources. Our competitive edge has always rested on our people — their skills, their adaptability, their ability to stay ahead of the curve. For the last thirty years, that edge has been real and hard-won. But if the tools that increasingly determine productivity, innovation, and economic output are owned, controlled, and potentially withheld by foreign powers, then our people's edge depends on someone else's generosity.

That is not an edge. That is a dependency.

"A Singapore that simply consumes AI built abroad is a Singapore that has outsourced its intelligence — and ultimately, its future — to someone else."

The Fable 5 shutdown was a preview. Not of a worst-case scenario — of an ordinary policy decision made for ordinary domestic reasons, with no particular malice toward Singapore or anyone else. The impact on everyone outside the US was incidental. That is precisely what makes it so alarming. We were collateral.

We are also not having the honest conversation about displacement. Some people have yet to come to terms with the fact that AI is going to bring about mass disruption to the way people work and live. There is no scenario in which that disruption does not happen. The only question is whether Singaporeans are the ones shaping and directing it — or the ones absorbing it. If we keep waiting, keep playing it safe, keep pretending this is a future problem, we will find ourselves on the wrong side of that question.

04What building sovereignty actually looks like

The European Commission's sovereignty package is instructive not because Singapore should copy it, but because of the thinking behind it. The EU is not trying to close itself off from American AI. It is trying to ensure it is not entirely dependent on it. The distinction matters: sovereignty is not about isolation, it is about having options.

For Singapore, the options look like this:

• Build and run AI on Singapore infrastructure — data that never leaves our borders, models that run on our compute, workflows that cannot be switched off by a foreign government.
• Invest in open-source models — capable, auditable, free from proprietary licensing, and increasingly competitive with closed-source alternatives.
• Develop home-grown platforms — not to replace every overseas tool, but to ensure that the core infrastructure of Singapore's AI capability is something we own.
• Train our people to build, not just consume — AI literacy that goes beyond using the tools that exist to building the tools that are needed.

These are not radical ideas. They are what every serious country is now doing. The EU is doing it at continental scale. France is requiring its civil servants to use homegrown video conferencing tools. Japan, South Korea, Canada — all of them are investing in AI infrastructure they control. Singapore needs to be in that conversation, not watching from the outside.

05What TrustedHub is doing about it

We are a 25-year-old company founded by Singaporeans in Singapore. We do not have the resources of a government or a hyperscaler. But we have spent two and a half decades building data infrastructure, and we understand what it means to own what you build.

Our Trusted ActionVerse platform is a home-grown, no-code agentic AI workflow builder — developed entirely in-house, running on our clients' own governed data, requiring no overseas API for its core function. Our inference cluster, currently under development, will enable capable open-source models to run locally — so that the organisations we work with are not dependent on any foreign provider's continued goodwill or governmental compliance.

We are also running the Hack to the Future Thinkathon series — live AI simulations in which autonomous agents are given operational control of entire organisations, compressed across simulated decades. Not to make a point about AI's potential, but to understand its failure modes before they arrive in the real world. Because the organisations that navigate this era best will not be the ones with the most powerful tools. They will be the ones who understand those tools deeply enough to govern them — and who built enough of their own infrastructure that no single foreign decision can take their capabilities offline overnight.

This is not a comfortable message. It is easier to keep using the best tools available and hope the geopolitical weather stays fair. But we have seen this week what happens when it does not. Singapore has built too much, over too many years, on too strong a foundation, to let that foundation be owned by someone else. The time to act on this is not after the next shutdown. It is now.