Every day, billions of pieces of content are published online: articles, videos, social media posts, responses generated by chatbots. The influx of digital information is no longer limited to the quantity of available sources. It now concerns the very nature of what we receive, with content produced or reformulated by automated systems whose mechanisms remain opaque to most readers.
Generative AI and Access to News: A Filter That Redistributes the Cards
One of the most significant recent changes in how we get information is the growing use of generative AI as an intermediary. According to the Reuters Institute Digital News Report 2025, published in June 2025, the weekly use of generative AI for news has increased over the past year, with a particular concentration among those under 25.
This shift alters the relationship with the source. The user no longer browses a list of results to choose an article. They receive a synthesized response from a system whose references they do not always see. The editorial sorting, once performed by the reader themselves or by a media outlet, is delegated to a text generation algorithm.
The data collected by Information Influx allows for mapping these changes in informational practices and better understanding the friction points between automation and real comprehension of the news.
However, the same report from the Reuters Institute points out a paradox: internet users trust journalistic sources more than chatbots to verify dubious information. Conversational assistants come in last among the means of verification cited by respondents. The adoption of a tool does not imply trust in its reliability.
European Regulation of Digital Platforms: What the DSA Changes in Practice
The Digital Services Act (DSA) requires very large online platforms to publish regular reports on their measures against misinformation. In practice, these transparency reports cover content moderation, algorithmic recommendation systems, and measures taken in response to systemic risks.
The European Commission has fined certain platforms for non-compliance. The case of X (formerly Twitter) illustrates the tensions between transparency obligations and resistance from private actors. These sanctions raise a fundamental question about the actual ability of regulators to compel companies whose resources far exceed those of regulatory authorities.
The DSA does not solve the problem of information overload, but it introduces a framework that requires platforms to document their algorithmic editorial choices. For the public, this means a beginning of visibility into the mechanisms that decide what they see and what they do not see.
Media Literacy and Verification: Practices That Resist Automation
In the face of the proliferation of content, manual verification reflexes remain the most reliable line of defense. Several concrete practices allow for actively filtering information rather than passively enduring the flow:
- Systematically cross-checking information with at least two recognized sources (investigative media, institutional websites, news agencies) before considering it reliable.
- Identifying the author and publication date of content, including for AI-generated responses, which do not always mention their references.
- Using specialized fact-checking sites, which remain more reliable than chatbots for resolving contested information.
- Being wary of content that provokes an immediate emotional reaction: virality often relies on outrage or surprise rather than factual rigor.
Digital media literacy is not only for students. Adult audiences, often less exposed to such training, are equally affected by confirmation biases and algorithmic bubbles that shape their news feed.
The Question of AI-Generated Images
The production of images by artificial intelligence has seen rapid growth in recent years. These visuals circulate on social media without systematic mention of their synthetic origin. Differentiating a real image from a generated image has become a major issue of visual literacy, particularly in contexts of political communication or event coverage.
Detection tools exist, but their effectiveness varies. Field feedback diverges on the ability of untrained users to identify these contents without technical assistance.
The Role of Traditional Media in the Face of Digital Disintermediation
Disintermediation, meaning direct public access to sources without going through a journalist, has been presented as a democratic advance. It also raises a problem of hierarchy. Without editorial work, a verified fact weighs as much as a rumor in an algorithmic news feed.
Traditional media retains an advantage that platforms do not easily replicate: the chain of editorial responsibility. A signed article engages its author, its editorial team, and sometimes a legal framework (right of reply, obligation to correct). Content generated by a chatbot engages no one.
This difference partly explains why, despite the decline in their direct audience, news media remain the first reference cited by internet users when they seek to verify a dubious fact, as confirmed by the Digital News Report 2025.
What Directions for Newsrooms
Several newsrooms are experimenting with the integration of AI tools into their processes, not to replace journalistic work, but to accelerate monitoring or transcription. AI in newsrooms serves as an assistant, not an editorial substitute. The boundary between technical assistance and delegation of editorial judgment remains a topic of debate within teams.
The influx of digital information will not slow down. The answers will come neither from a single tool nor from regulation alone, but from a combination of legal frameworks, individual verification practices, and editorial responsibility maintained by the media. The ability to inform oneself correctly depends less on the quantity of accessible data than on the quality of the filters, whether human or institutional, that each chooses to interpose.
