Disinformation, AI, Platforms: Who Controls the Digital World?

The challenges are numerous; however, I will mention only three.

The first is reconnecting individuals with reality. The digital world is not separate: its effects are very real. An online rumor, digital harassment, or foreign interference in an election all have concrete consequences. Yet these harms are often trivialized because they occur on screens.

The second challenge is pace. In the digital space, everything is instantaneous, viral, and amplified. Content spreads globally within minutes, while educational, legal, and institutional responses evolve much more slowly.

Finally, the issue of informational sovereignty must not be underestimated. An increasing share of public debate takes place on private platforms, often foreign, which determine what is visible, amplified, or invisible. This directly affects democracy and trust in information.

Generative AI now makes it possible to create highly realistic texts, images, voices, or videos within minutes. Disinformation has also become profitable: it spreads easily through shocking or emotional content, which travels faster and reaches ever larger audiences.

Algorithms do not prioritize truth, but engagement. Added to this are our cognitive biases: we are more likely to believe information that confirms our opinions. Finally, traditional editorial filters have largely been replaced by systems designed to capture attention rather than ensure reliability.

This is precisely the value of an interdisciplinary approach such as the “CAS en Gouvernance de l’information et confiance numérique” which I co-direct. It allows disinformation to be analyzed as a global system rather than a simple individual deviation.

AI-based detection systems capable of analyzing inconsistencies in images, voices, movements, or metadata have been developed. In addition, forensic techniques make it possible to refine these analyses.

However, we must remain clear-eyed: detection often occurs at the end of the chain. False information has already been seen and shared before corrections can be made. Even when debunked, it can leave lasting traces.

Therefore, the key challenge lies more in certifying the origin of content than in debunking it afterward. These technical and legal aspects are addressed in the course “Criminal Law Mechanisms and Digital Evidence”, which I designed and teach.

 
Absolutely. However, I prefer to speak of legal responsibility. Twenty years of self-regulation have shown their limits. These platforms are not mere hosts: their algorithms make editorial choices on a massive scale. They decide what is highlighted, hidden, amplified, or distorted.

When a teenager is exposed to harmful content, it is not an accident: it is the result of a system designed to capture attention. The same applies when AI enables the creation of pornographic deepfakes: the problem is not only individual misuse, but the architecture that makes it possible.

The European Union has begun to regulate these practices; however, much work remains. In Switzerland, certain digital offenses already exist, but platforms still operate with limited safeguards. These issues are central to the course “Régulation des plateformes de réseaux sociaux dans une société démocratique” which examines these power dynamics in detail.