There is an old joke in Brussels, repeated with more affection than it deserves: Europe does not build the future, but drafts the amendments.
It was once said as a jab, about a continent too slow to deliver innovation and good only at drafting laws.
Silicon Valley writes the code; China builds the models; Europe writes the rules. This, more or less, is how the division of roles in the global race for artificial intelligence has been described.
But with the new European Artificial Intelligence Act, the main rules of which began to apply on August 2, it seems that this time Europe is getting ahead of the future, rather than following behind it.
The Artificial Intelligence Act is not a fence against innovation, but the first serious attempt to get ahead of the harm.
That is no small achievement, in fact.
And it does not belong only to the 27 countries that voted for it. Every capital on this continent has an interest in ensuring that this effort does not fail.
A generation raised inside the smartphone, the one Jonathan Haidt calls the “anxious generation,” is the “guinea pig” of the largest uncontrolled experiment ever conducted on human attention.
And the data pouring in every day leave little room for doubt: the rise in anxiety, depression and self-harm among adolescents almost exactly coincides with the years when the phone in our pocket went from being a phone we used to talk to, to a phone that lives with us.
We can debate the numbers of the studies and how they were conducted as much as we like. But no serious person doubts anymore that technology has gradually enslaved us all.
We conducted the experiment on our children before the regulatory reflex was born, which has now become law in Europe.
What happened with social networks and their impact on our lives was only the warning.
Artificial intelligence is the same experiment, only much faster and this time on everyone, not just on young people.
But this time Europe seems to have a choice that it did not have with social networks: to establish rules before the damage becomes a fortress.
This is what truly matters about the European Artificial Intelligence Act. It does not attempt to regulate every use of AI in the same way. It starts from risk. Some practices are banned. High-risk systems, used in areas such as education, employment, essential services or law enforcement, are subject to much stronger obligations. And, since this scorching August, transparency rules require certain content created or manipulated by AI to be identifiable, including deepfakes.
It may be a law that is not quite proportionate to a technology that moves faster than our ability to understand it. It will be changed, challenged and, most likely, technology will invent something else that the law today still does not know how to name.
But Europe has established an important principle this time:
Not everything that technology can do should be allowed by humans.
This may be the most exportable idea Europe has produced this decade. In a race where others measure the power of technology, Europe is trying to set its boundaries. Technological innovations do not exclude responsibility, and human rights do not get suspended simply because technology moves faster than the law.
The AI Act, European rules on data, and the framework for platform accountability will define the digital environment in which Europe’s teenagers will grow up, including ours.
You do not need a membership certificate to protect a fourteen-year-old girl in Tirana from the same digital environment in which a fourteen-year-old girl in Paris or Berlin lives.
AI can help a young person learn a language, solve math exercises, or create something they previously would not have been able to create. But it can also tell them, with complete confidence, something that is untrue.
Young people do not need to fear AI, but they can use it without losing the habit of thinking for themselves.
The fourteen-year-old girl whose well-being we are concerned about today is tomorrow’s voter. Her ability to distinguish a real image from a fabricated one, journalism from propaganda, persuasion from manipulation, will also determine the quality of the democracy she will inherit.
Knowing how to ask a chatbot the right question is not enough. You must also know when to question its answer, why an algorithm brought exactly that content before your eyes, whether an image is real, and whether your privacy is safe.
As for our part of the world, we can start with something much simpler: applying the protective logic of the AI Act right now where children are concerned.
Countries that are still waiting at Europe’s door have more reason than anyone else not to wait for permission to protect their own. After all, the time for waiting has passed.
