We need to stop talking about AI governance as if we are simply waiting for a seat at someone else’s table.

For the past two years, the narrative has been dominated by the North. We’ve seen the EU AI Act, the various voluntary commitments in the United States, and the debates in Beijing. And while those models are intellectually interesting, they are increasingly being exported to the Global South as if they were universal truths.

But here is the reality that isn’t being said out loud: Importing regulatory frameworks is not a strategy. It’s an admission of defeat.

When we simply copy and paste policies from Brussels or Washington, we are ignoring the fundamental economic and social realities of the Global South. We are importing software that was never designed to run on our hardware.

My name is Nafiul Ahmad Rafi, and today, I want to talk about why we need to move from ‘regulatory consumers’ to ‘sovereign architects.’

The first trap we are falling into is the belief that ‘safety’ looks the same everywhere. In a high-income nation, AI governance focuses on mitigating the displacement of white-collar workers, or managing the risks of high-compute models that process vast amounts of digital wealth.

But look at the Global South. Our challenges are different. We aren’t just worried about AI bias in a chat application; we are worried about AI infrastructure being built on top of data extraction. We are worried about our local languages being erased because the foundational models weren’t trained on our vernaculars. We are worried about energy usage.

If we don’t build our own frameworks, we are essentially allowing the global tech giants to define the rules of our own digital sovereignty.

So, what does a sovereign AI framework actually look like?

It starts with three non-negotiables.

First: Data Sovereignty as a Tradeable Asset. For too long, the Global South has been the training ground for Northern AI. Our public records, our local knowledge, and our economic data have been harvested without compensation. A sovereign governance framework mandates that our data stays local, that it is treated as a collective asset, and that we have the power to decide who uses it and for what purpose. We shouldn’t be donating the raw materials of the future—we should be building the industry here.

Second: Infrastructure as Governance. You cannot regulate what you do not own. If we are completely dependent on cloud clusters located in the North, our governance is merely a suggestion. It is a polite request. We need to prioritize regional, state-backed, or public-private GPU clusters. We need to invest in energy-efficient data centers that serve our local markets first. If we want our policies to have teeth, we need the compute power to back them up.

Third: Contextual Responsiveness. An AI model that works in a San Francisco office might be completely dangerous in a rural clinic in our region. It lacks the context. It lacks the history. It lacks the nuance. Governance in the Global South must mandate that any AI deployed in our public services—whether that’s healthcare, education, or agriculture—must undergo ‘contextual validation.’ It must be tested against our social norms, our climate realities, and our specific needs. If the model doesn’t understand us, it doesn’t belong here.

I know what the skeptics will say. They’ll say, ‘We don’t have the time to build this. We need to catch up. Why reinvent the wheel?’

But the wheel is being reinvented. We are at the very beginning of the AI era. If we establish our own standards now—standards that prioritize development, equity, and local agency—we won’t just be ‘catching up.’ We will be setting the pace for the rest of the world.

Think about the mobile payment revolutions we led. We didn’t wait for the West to show us how to digitize our economies; we built the stack from the ground up because we needed it. We did it better, we did it faster, and we did it at a scale they couldn’t even imagine.

AI governance is the next frontier.

To the policymakers, to the tech founders, and to the researchers across the Global South: stop asking for permission to define your future. Stop checking to see if your policies align with the EU or the US.

Ask yourself: Does this policy serve our people? Does it protect our sovereignty? Does it build our capacity?

If it doesn’t, throw it out.

We have the demographic dividend. We have the data. We have the ingenuity. The only thing we have lacked so far is the confidence to say that our way is not just ‘different’—it is necessary.

The future of AI is not a monolith. It is not going to be determined by a single regulatory body in a single city. It is going to be determined by the regions that have the courage to build their own guardrails, their own infrastructure, and their own future.

It’s time to start building.

I’m Nafiul Ahmad Rafi. And if you’re ready to move beyond the imported narratives, let’s get to work. Reach out, let’s start the conversation, and let’s define what the next decade of sovereign AI looks like, together.