AI Needs Data. Why Doesn’t It Pay for It?

I had the pleasure of moderating a workshop at the Ambassade de France en Allemagne dedicated to developping a new economic model for media during the seminar “Media and Disinformation”, in the presence of Gérard Larcher, President of the French Senat.
AI models need four things: talent, computing power, energy and data. The first three are paid for. The fourth is still too often used without clear authorisation, sufficient transparency or fair remuneration.

Why? Why should journalistic content, books, music, images, films, creative works and quality information be treated as free raw material?
Journalism Is Not Just Data. Journalism is a strategic asset.

It requires time, investigations, editorial teams, correspondents, fact-checking, fieldwork and sometimes risk. It produces reliable, structured and contextualised information. And precisely for that reason, it has immense value for AI systems. If AI needs reliable information to function, then it must also contribute to financing the ecosystem that produces that information.

This Is Not Innovation Against the Media
The real issue is value sharing. Europe already attempted a first response with the neighbouring rights of press publishers.

But generative AI changes the scale of the problem: content is no longer used only to reference information. It is used to train, enrich and feed systems capable of answering users directly.
Media outlets produce the information. AI assistants deliver it. Publishers gradually become invisible.

A Democratic and Economic Challenge
Without a viable economic model for the media, there will be fewer investigations, less pluralism, less local journalism and fewer checks and balances. This is therefore not only an intellectual property issue. It is a democratic one.