AI Act: Europe Balances Regulation and Technological Dependence

Europe Faces the Challenge of AI Regulation

As the world turns towards the large-scale development and deployment of artificial intelligence, Europe has chosen to legislate. The recently adopted AI Act aims to regulate the uses of AI, but its effectiveness is already being questioned. By focusing on regulation without a clear industrial strategy, Europe risks becoming dependent on AI giants.

The United States is investing hundreds of billions in AI infrastructure, while China is training millions of engineers. Meanwhile, Europe is busy producing hundreds of pieces of legislation. The AI Act, although a regulatory advancement, could quickly become obsolete in the face of rapid technological evolution. This phenomenon illustrates the "Collingridge Dilemma," theorized in 1980, which highlights the difficulty of regulating an emerging technology before its impacts are fully understood.

Industrial Urgency Against Regulatory Mirage

The AI Act relies on a risk-based approach, a relevant strategy for prohibiting abuses such as manipulation or widespread biometric surveillance. However, it misses the true nature of this revolution: artificial intelligence is no longer just a software tool; it is a heavy industry of transformation, not merely a technological evolution of digitization.

The idea that regulation protects Europe is a delusion. As Arthur Mensch, co-founder and CEO of Mistral AI, analyzed during his Senate hearing on May 12, the heavy regulatory burden in Europe creates an entry cost that only large companies—and consequently American giants—can absorb due to their colossal resources and lobbying capabilities.

While some view this as an opportunity for local players to stand out, this regulation stifles the ecosystem and local startups. Unlike the United States, Europe imposes the juggling of multiple labor laws, tax systems, and languages (key elements of our wealth and identity). When viewed on a global market scale, this situation significantly hinders the scaling of our European companies.

In its example and with the trajectory already set for 2030, if 10% of the productivity gained at the European level relies on American models, Europe will effectively pay them a massive technological rent while having to bear the associated mass job destruction (Mensch estimates this value transfer at several hundred billion annually by 2030). Without reinvestment in AI, it will deepen its trade deficit and create no leverage in its trade relations.

Currently, American giants are deploying hundreds of billions of dollars to monopolize access to components, infrastructure, and energy, thereby seizing future capacity and reserving the associated resources to support the infrastructures that will enable these profits.

For Europe, the real barrier is no longer regulatory or software-related, but rather energetic and material. Faced with extra-European actors fueling a demand that exceeds supply, they generate a supply crisis where demand far exceeds global production capacities in chips, servers, and electricity.

Open Source: A Lever for European Sovereignty

In this global race, Europe has a major asset, often underestimated in regulatory debates: Open Source. Beyond a software distribution model, it is a guarantee of transparency and a decentralized engine of innovation to combat an oligopoly.

Open Source is the only field where a mid-sized European company can today build a sovereign AI. But we must be realistic: the leading open reference models are currently primarily Chinese, such as Kimi, Qwen, DeepSeek, and Mimo, ahead of American and European counterparts.

If French players like Mistral AI or LightOn manage to exist alongside American titans, it is largely due to a strategy of open models and the use of open source in the foundations of AI platforms and agents. They have paved the way in Europe, but the continent has yet to create the ecosystem to support and scale them to the level of their competitors.

In reality, digital sovereignty goes far beyond Open Source and equally depends on cloud services, semiconductors, network infrastructures, cybersecurity, skills, public procurement, funding, and industrial capacity on a global scale.

The Specter of a Regulatory Rather than Innovative Europe

Legislating in hopes of defending against American monopolies does not work. Without a strong industrial strategy to support the sector, the risk is real: slowing the rise of local players while allowing foreign competitors to develop in a more flexible framework before entering the European market.

European companies are already paying the price: compliance costs crush SMEs while Big Tech absorbs them without blinking and continues their relentless lobbying. Additionally, there is the exodus of our engineers and researchers, who leave ecosystems where experimentation is stifled by bureaucracy without financial recognition that meets international standards.

And lacking models trained on our data, we are already inheriting the cognitive and linguistic standards of those who produce them, resulting in: impoverishment of language, standardized thinking, and an inability to break free from a framework that will be imposed and continuously disseminated.

For a Living Governance

Rather than a fixed text like the GDPR, artificial intelligence requires the construction of a living governance and a flexible framework that serves our society.

First, we must strengthen safeguards where risks become clearer: deep fakes, critical and ideological biases, as well as censorship. Transparency regarding training data and model biases must become a selection criterion, not a regulatory constraint.

Economically, three measures would be sufficient to change the trajectory. First, index the thresholds of the AI Act to a technical committee with annual reviews rather than the European legislative cycle. Next, direct a portion of public procurement towards European open source AI solutions as an indirect funding lever for R&D. Finally, address AI sovereignty through industrial consortiums, not through laws or directives.

The Collingridge Dilemma is not a fatality; in the past, we Europeans have managed to build and regulate simultaneously, as seen in nuclear energy or aerospace.

Europe must choose: to endure the innovation of others within a framework it alone will have respected, or to reclaim its place in the race. Without alignment of regulation and the ability to generate profitable growth, reinvesting the margin gained into R&D and creating strategic interdependencies within our European ecosystem, we will remain vassals of empires.