Talking about AI literacy before tech sovereignty
A country cannot claim technological sovereignty if its people lack the skills to understand and use AI, writes Kenneth Brincat
On June 3 the European Commission unveiled its Technological Sovereignty Package, an important step towards greater European control over the technologies that run critical systems, from health solutions to energy grids, from public services to infrastructure. It reaches those sectors, but also the technologies beneath them, from cloud storage to AI infrastructure.
It is ambitious: the Cloud and AI Development Act, a revised Chips Act, the EU Open Source Strategy, a drive to build AI gigafactories, and a tiered system of sovereignty assurance levels for public-sector workloads. The figures attached run to hundreds of billions of euros over a horizon stretching past 2035 – because that is what sovereignty actually costs: data centres, computational resources, semiconductors, sustained across a decade of investment.
Malta is not at the periphery of these discussions. We are focused on how the country plugs into European compute capacity, where the gigafactories are located, and how the new assurance levels should shape what the public sector procures – questions addressed across our national strategies: Malta Diġitali, the National AI Strategy, the Malta Digital Decade Strategic Roadmap and others.
It is why Malta was selected to host an EU AI Factory Antenna, CALYPSO, led by the MDIA and linked to Greece’s PHAROS AI Factory, connecting local startups, SMEs, researchers and public bodies to Europe’s most advanced supercomputing. Far from stepping off the European track, Malta is wiring itself into it. What remains is to address technological sovereignty head-on, as a national strategy in its own right and in conversation with stakeholders, so that our policy direction maximises integration with the European effort as a whole.
But it is easy to miss the wood for the trees. Technological sovereignty cannot be meaningfully achieved without a broad base of AI literacy across society, because sovereignty is not one thing but three. There is data sovereignty, under whose law information sits and who can compel access to it. There is technological sovereignty, who owns and builds the cloud, the chips, the models. And there is digital capability, whether a population can actually use, and question, the tools that now shape its economy.
Much of the recent criticism collapses these into a single word and then judges a capability programme as though it were a procurement decision about the national stack. That is where the analysis goes astray.
‘AI for All’ addresses the third pillar. It is a free literacy course, built with the University of Malta and launched by the MDIA, open to residents through their eID, after which a participant may choose a year of ChatGPT Plus or Microsoft Copilot. The aim, stated and designed for, is capability: that a nurse, a shopkeeper or a sixth former can use these systems competently and understand their limits. The subscription is a downstream consumer benefit attached to that aim, not a procurement decision about the national technology stack.
Does putting non-EU tools into people’s hands cut against sovereignty? The tools are already here. Maltese students, civil servants and businesses use them daily, on personal accounts governed by no one – Malta ranks third in the EU for generative-AI use, with nearly half of adults already using these systems, according to Eurostat’s 2025 figures.
Whatever dependency exists, the state did not create it; it stepped into one already in the room. And by stepping in, it gains what a million individual clicks on ‘accept’ never provided: a single point of administration, the discipline of public oversight, and the leverage of a sovereign counterparty to insist on terms – EU data residency, no training on users’ content, audit rights. The honest question is not whether the state should be involved, but whether those terms are strong enough. That is something to negotiate and to scrutinise, not a reason to refuse.
There is a further gain the language of extraction misses. Because the programme runs through the MDIA and the national eID, it produces something Malta has never had: an aggregate, anonymised picture of how the country is actually adopting artificial intelligence – who enrols, in which sectors, with what completion rates, and which tools they choose. This reads no one’s prompts; it is population-level adoption data that policymakers have until now been left to guess at.
Where the critique sees only data flowing out of the country, there is a flow the other way. For the first time the state can see its own digital transformation rather than infer it – and a state that can measure its readiness is far better placed to steer it. In the absence of the sovereign infrastructure we cannot yet build, that visibility is not a trivial consolation. It is itself a form of sovereignty: the capacity to know, and therefore to govern, one’s own digital trajectory.
Nor is any of this a departure from the European framework, it is an enactment of it. Malta was an early mover to implement the EU AI Act in domestic law, through the 2025 legal notices that gave the MDIA and the Information and Data Protection Commissioner their supervisory roles; the safeguards meant to protect the citizen are already in place.
And the Act itself recognises that consumer protection depends on an AI-literate public: it carries an AI-literacy obligation, a requirement that those using and deploying these systems actually understand them. Through AI for All, Malta is operationalising a provision of the very framework invoked against it – among the first EU states to act on it in practice.
We are rightly asked to approach all this with objectivity, and to resist blind optimism. Objectivity means holding two things at once. Technological sovereignty is a value worth striving for and defending – a matter of political economy and industrial strategy, of contracts, compute and capacity – and Malta must pursue it in earnest.
But sovereignty is more than infrastructure. It also depends on whether citizens, businesses and public institutions can understand, use and question the systems that increasingly shape their lives. AI literacy is not secondary to technological sovereignty. It is one of its foundations.

Kenneth Brincat is a doctoral researcher in digital innovation readiness. He is also chief executive officer of the Malta Digital Innovation Authority, which administers the AI for All programme.