The great bluff

Posted on: 25 November 2025

There is a precise moment when a dominant narrative begins to crumble, and that moment arrived this week in the world of artificial intelligence regulation. Everyone is talking about the "great race": the United States accelerating to beat China, Europe seeking to protect its citizens, Big Tech pushing for deregulation. It's a reassuring story in its clarity: defined actors, clear interests, coherent strategies. The problem is that it isn't true.

American chaos

Take Donald Trump and his executive order that should block all state laws on artificial intelligence in the USA. On paper it's an aggressive and coherent move: federal standardisation, elimination of the patchwork of 50 different regulations, green light for innovation to compete with China. It should be the classic show of force from an administration that knows what it wants and has the means to achieve it. Instead, what happens? The governor of Florida Ron DeSantis, Republican and natural ally, calls it a "subsidy to Big Tech" and "unacceptable federal overreach". Sarah Huckabee Sanders in Arkansas aligns. Steve Bannon and his associates define the operation as a betrayal of American interests in favour of Silicon Valley billionaires. The executive order, which should have been signed last Friday, has been put on hold. Not because of Democratic opposition, but due to chaos within their own camp.

European civil war

Let's move to Europe, where the standard narrative wants a European Union proud of its standards: GDPR as a banner, the AI Act as a declaration of digital sovereignty, the famous "Brussels effect" that influences the entire world. Last Wednesday the Commission presented the "Digital Omnibus", a package of reforms that softens both data protection and AI regulation. The official justification is "simplification for small businesses". The reality is that it would allow companies to use personal data to train AI models on the basis of "legitimate interest" without explicit consent, and would delay by a year the rules on high-risk systems. Max Schrems, the Austrian lawyer who has spent a decade fighting legal battles against Facebook and Google, calls it "the biggest attack on European digital rights in history". European parliamentarian Brando Benifei and dozens of civil society organisations are in open revolt against their own Commission. It's not an elegant and strategic capitulation, it's an institutional civil war where part of the European apparatus is literally trying to "override everyone else in Brussels", using Schrems' words.

Work disappearing faster than solutions

But there's another dimension where the illusion of control manifests itself even more brutally: the world of work. The official narrative promises that AI will create more jobs than it destroys, and that public intervention, through retraining and training programmes, can manage the transition. It's a reassuring story that presupposes the existence of an institutional safety net capable of operating at the same speed with which technology eliminates entire professional categories. The problem is that this net doesn't exist. A report from June 2025 by the Financial Services Union shows that almost 90% of workers in the financial sector believe that AI will cause significant job displacement, with some experts predicting the replacement of almost half of all entry-level roles in the technology, financial, legal and consulting sectors. This isn't happening in a hypothetical future, it's happening now. Redundancies in back office services, basic code writing, customer assistance are rapid and concentrated. Meanwhile, job centres and professional training systems operate with budgets, structures and decision-making speeds designed for a twentieth-century economy. It's like trying to put out a fire with a bucket whilst someone continues to pour petrol.

Finance out of control

And then there's finance. Whilst politicians and regulators debate who should control what, Michael Barr, Vice Chair of the Federal Reserve, warns that artificial intelligence algorithms are creating "herding behaviour" in financial markets that could amplify volatility in unpredictable ways. The Bank for International Settlements confirms that over 70% of global equity transactions now involve algorithmic components. Recent studies show that machine learning models, especially those based on reinforcement learning, can spontaneously develop behaviours that resemble cartels: collusive coordination without explicit programming. The SEC and CFTC know all this. They're developing guidelines. But the system continues to accelerate because no financial institution can afford to slow down unilaterally whilst competitors keep the accelerator pressed. It's the classic prisoner's dilemma in real time, but with algorithms that operate at speeds that exceed human capacity for intervention. The flash crash of 2010 lasted minutes. The next one could last days, and this time the interruption mechanisms might arrive too late.

The physical race nobody sees

There's then an even more concrete dimension that escapes most analyses on regulation: the physical reality of artificial intelligence. Whilst everyone discusses laws, privacy and algorithms, the real race is taking place on much more material terrain. Training the most advanced models requires a massive quantity of Nvidia chips, which have become rarer and more precious than gold. It requires data centres that consume electricity at levels that are beginning to put pressure on national grids. The narrative of a "democratic" and "accessible to all" AI clashes with the brutal reality of physical bottlenecks. Who controls the most advanced chips? Who has access to the energy necessary to power these systems? These questions are far more determinative than any parliamentary debate on algorithmic transparency. And here the illusion of control becomes even more evident: regulators concentrate on software, on policy, on digital rights, whilst real power consolidates at the level of physical infrastructure. It's like regulating air traffic whilst someone else controls the fuel supplies and decides who can fly and who cannot.

The illusion of control

Here's where it becomes interesting: all these actors are losing control simultaneously, but nobody can admit it because the admission itself would cause the panic they're trying to prevent. Trump must project strength and strategic clarity, "we'll beat China!", even whilst his own party fractures on the question. The European Union must maintain the façade of "protector of digital rights" even whilst capitulating under economic pressures it can no longer ignore. Financial authorities must appear in control even whilst observing systems that could collapse more quickly and last longer than any previous crisis. Governments promise retraining programmes whilst workers lose their jobs faster than the programmes can even be designed.

The complex system gone mad

The result isn't a great strategic game between world powers. It's something far more dangerous: a complex system where each actor reacts to the movements of others with increasingly rapid and increasingly uninformed decision cycles. Trump observes China accelerating and decides to deregulate. Europe observes Trump deregulating and decides to loosen its own rules so as not to be cut out. China observes both and accelerates further. Meanwhile financial markets operate at speeds where an algorithm can cause billions in damage before a human being has finished reading the first alert. In complex systems theory, this is the exact moment before a completely unexpected behaviour emerges. Not because someone is planning something evil, but because the system itself has exceeded the collective cognitive capacity of those who should be managing it.

So what should we really be asking ourselves?

Here the real question is: does anyone remain at the wheel, or are we all watching a car that's accelerating by itself whilst the passengers argue about who should have the steering wheel? And this, perhaps, is the reason why this narrative is the most dangerous to contemplate. Because admitting that nobody is really in control means accepting that our institutions, democratic, financial, regulatory, could be structurally inadequate for the type of complexity we have created. It's far more reassuring to believe in the chess game, even when all the evidence suggests we're playing something completely different.

The only certainty is uncertainty

Next week Trump might still sign that executive order. Or he might not. Europe might approve the Digital Omnibus. Or it might back down under parliamentary pressure. Financial markets might continue to function perfectly. Or we might wake up one morning with a crash that this time lasts days instead of minutes. Thousands more workers will lose their jobs whilst politicians announce new training programmes that will arrive too late. Nvidia will continue to control the physical bottleneck whilst everyone discusses algorithmic ethics. The only certain thing is that nobody really knows what will happen. And this, in a world that has built its security on the illusion of control, is the most destabilising fact of all.