When a newsroom chooses to cover a news item rather than a parliamentary report, it is not always an editorial choice. It is often an economic arbitration: the news item generates more clicks, thus more advertising revenue. This mechanism, seemingly mundane, alters what we consider important news.
Advertising Model and Online News Selection
The majority of online media derive their revenue from programmatic advertising. Specifically, each page view triggers an automatic auction among advertisers. The more visitors an article attracts, the more it earns.
This leads to a direct effect on the editorial line: high click-potential topics take precedence over public interest subjects. A controversial headline about a political statement taken out of context will generate more revenue than an in-depth investigation into a bill. Newsrooms are aware of this, and those that rely solely on advertising do not always have a choice.
This invisible sorting is not the result of organized manipulation. It stems from a market logic where the reader’s attention is the product sold to advertisers. One can delve deeper into the mechanics of the business on Contre Informations, which details how these financial circuits structure content production.
The problem is exacerbated by 24-hour news channels, which also operate on an instant audience model. The result: an overrepresentation of anxiety-inducing events and controversies, to the detriment of complex topics that require time to read.
Platform Algorithms and Information Bubble
Most French people no longer seek news directly on a newspaper’s website. They receive it through a feed: Facebook feed, YouTube recommendations, Instagram stories, TikTok suggestions. Platforms and social networks have become the primary point of contact with information, especially among young people.
The algorithms of these platforms do not rank content by journalistic relevance. They rank by likelihood of engagement: time spent on the post, comments, shares. A nuanced article on an economic reform goes under the radar. A 45-second video with an alarmist title circulates for days.
What Algorithms Favor in Practice
- Content that triggers an emotional reaction (anger, outrage, surprise), because it generates more interactions than factual analyses
- Short and visual formats, compatible with rapid scrolling, which pushes newsrooms to simplify topics that cannot support it
- Sources that the user already consults, reinforcing the information bubble and reducing exposure to contradictory viewpoints
This algorithmic filtering creates a paradox: we have access to more sources of information than ever, but we see an increasingly narrow slice of the news. Digital has diversified the offering, but not necessarily the actual consumption.
Trust in Sources and New Information Intermediaries
Trust in traditional media is eroding. Institutional communication practices, factual errors repeated endlessly, and the perception of a biased editorial line fuel this distrust among a significant portion of the French public.
At the same time, new channels are taking over. Content creators on YouTube or Twitch cover the news with their own codes. These formats are appealing because they give an impression of closeness and authenticity that traditional newsrooms struggle to replicate.
AI Chatbots as News Sources
According to data from the Reuters Institute published in July 2026, 10% of global news consumers use AI chatbots for information. The trust placed in these responses remains significantly lower than that given to traditional media.
This figure may seem modest, but it signals a structural change. A chatbot does not produce information: it reformulates and aggregates existing sources, often without citing them. The reader receives a smooth response, without knowing whether it comes from journalistic investigation or an opinion piece.
The European Union has begun to regulate this gray area. The European AI Act has made it mandatory, effective August 2, 2026, to provide explicit labeling for chatbots, deepfakes, and AI-generated informational texts without human editorial review. All synthetic content intended for information must now be identified as such.
Attention Economy and Reading Reflexes
At the end of the chain, our own reading behavior fuels the system. We scroll quickly, read headlines, and share before reading the article. These reflexes are not an individual flaw: they are the logical response to an information-saturated environment.
Several concrete mechanisms amplify this dynamic:
- Push notifications from news apps prioritize breaking news, which trains the brain to retain only urgency
- The “headline” format of aggregators (Google News, Apple News) selects titles based on freshness and volume of publication, not editorial quality
- The perceived free nature of online information devalues journalistic work and makes it more difficult to fund long investigations
The result: a fragmented, emotional, and moment-centered perception of the news. In-depth topics, those that require context and perspective, disappear from the majority of readers’ field of vision.
This cycle between economic model, algorithms, and reading habits will not be broken by individual will alone. Transparency about media funding, the obligation to label AI-generated content, and the diversification of sources remain the most concrete levers to change the game. We are only at the beginning of this regulation.
