Malta, Greece and Estonia may offer a glimpse of what the future could look like when governments trade access to their data for privileged access to artificial intelligence.
All three are EU members and subject to the bloc’s privacy rules under the AI Act. Yet all three have reached agreements with Sam Altman’s OpenAI. Malta’s deal, announced quietly in May, gives citizens free access to ChatGPT Plus for a year, along with training on how to use AI more effectively.
In return, the Maltese government makes institutional data — as well as citizens’ data unless they explicitly opt out — available to OpenAI. Agreements with Greece and Estonia are somewhat more limited, but follow a similar model. Governments provide anonymised data about public services and citizens in exchange for preferential access to AI tools, including for education.
The United Arab Emirates is pursuing a similar approach through its Stargate project.
The underlying bargain is striking: sovereign governments are giving a technology company access to information they would never normally share with another state, in exchange for subsidised or free AI services. It is a powerful example of how influence may be exercised in the 21st century.
The hunger for data
The reason for the exchange is straightforward. The biggest AI companies have already trained their models on much of the publicly available information on the internet. Their next challenge is finding new data to make their systems more capable.
Every time someone uploads a medical report to Gemini to understand it, Google potentially gains useful information. Every time a user asks Claude about symptoms or seeks an analysis of a stock, Anthropic gains another piece of data.
Behind these systems is also a growing army of human workers who review and label material before it is used to train AI models. Companies such as Sama, Alignerr, Outlier AI and OpenTrain AI employ workers to identify offensive, violent, explicit and otherwise controversial content.
Much of this work is outsourced to lower-cost countries with large English-speaking populations, including parts of East Africa, India and the Philippines, and even refugee communities in Lebanon. Similar workers also operate largely out of sight in Italy.
But the relationship between humans and AI is changing in another way, too.
When AI becomes personal
AI is increasingly moving beyond work and productivity into people’s emotional lives.
Research suggests that a significant number of young Americans are already using AI for companionship. Some teenagers have developed romantic relationships with chatbots, while some people who have real-life partners also maintain secret relationships with AI systems.
The phenomenon has created a new market for applications that offer the appearance of friendship, therapy, grief support, romance or sexual companionship.
Replika, for example, can create a digital personality based on a user’s messages and other material. Its business model shows how emotional dependence can also become a subscription service: users may begin by interacting with an AI as a friend, but more intimate features can require a paid upgrade.
When Replika temporarily restricted its romantic functions in 2023, some users described the experience as a form of grief or the sudden end of a relationship.
The issue, however, is bigger than the psychology of individual users. It is about power.
Why Acemoglu matters
This is where economist Daron Acemoglu enters the picture.
The Nobel Prize-winning MIT professor, one of the world’s most prominent economists studying the relationship between technology and society, has become an outspoken critic of the way AI is being governed — or, in his view, barely governed — in the United States.
Acemoglu argues that leaving the development of AI largely in the hands of a small group of powerful technology executives could have serious consequences for employment and society.
His criticism is particularly significant because he is hardly an outsider to the capitalist system he is questioning. He is a highly respected economist with a strong quantitative background and a position at one of America’s leading universities.
Some observers have interpreted recent criticism of Acemoglu as part of a wider backlash against his position on AI. The argument is that the debate is not simply between people who support technological progress and those who oppose it, but between different visions of what technological progress should look like.
That makes the question of who controls AI increasingly difficult to ignore.
A $23 trillion bet
The scale of the financial stakes is enormous.
Anthropic is reportedly heading towards a valuation of around $2 trillion, while OpenAI could soon follow with a valuation of at least $1 trillion. The seven major Wall Street companies most heavily exposed to AI are spending hundreds of billions of dollars each year on the technology, increasingly using debt to finance those investments.
Together, those companies are worth around $23 trillion on the stock market — a valuation that already reflects enormous expectations about the future of AI.
Companies linked to the AI ecosystem, from chipmakers to businesses building products around AI models, account for roughly 45 per cent of the $69 trillion value of America’s main stock-market index.
That leaves the world facing a paradox.
AI may become one of the most transformative technologies in modern history, but its development is increasingly controlled by a small number of extraordinarily powerful individuals and companies.
If even part of the promises surrounding the AI revolution fail to materialise, the financial consequences could be enormous — potentially far greater than the collapse of Lehman Brothers in 2008.
For now, many investors and governments appear willing to accept the concentration of power in Silicon Valley as the price of technological progress.
The real question is whether the world should.
Be First to Comment