A quiet systemic shift: how AI dependency takes root

The language models most used by children and teenagers, such as ChatGPT, Gemini, DeepSeek and Copilot, are operated by companies headquartered in the United States or China. They follow the rules of their home countries, not those of the people using them. Conversations are processed on foreign servers and can be used there to improve the models.

That is a wake-up call to redefine the role of data protection. If it is as important as we claim, we would have to forbid AI for children. But prohibition has rarely helped. The real question is how we use AI, starting in childhood.

In February 2026 it was reported that the US government instructed its diplomats to argue against digital sovereignty measures in EU countries (Reuters). Every country promotes its own industry. Whoever establishes infrastructure as a standard early secures long-term market access and influence, and with AI that happens faster and deeper than with social networks.

The resulting dependency is not technical, it is cognitive. Whoever learns early to outsource complex questions to a system develops different problem-solving patterns than someone who works through the difficulty alone. This is not a judgement of the technology. It is what happens when a powerful tool is used without a framework.

The same holds for adults who lose their jobs: those who only learned to consume AI are replaced by it, those who learned to think with it are hired. Teaching that difference is why we exist.