Reliable information refers to content whose source is identifiable, the production method is transparent, and the facts are verifiable. In the digital age, this operational definition becomes a concrete issue: dissemination channels are multiplying, intermediaries are changing in nature, and the boundary between human-produced content and machine-generated content is blurring. Understanding the mechanisms that ensure or undermine the reliability of information allows for better navigation in this environment.
AI-generated content and labeling obligation: what the AI Act changes
Since August 2, 2026, Article 50 of the European AI Act requires AI system providers to make the artificial or manipulated nature of texts, images, sounds, and videos identifiable in a machine-readable format. Systems already deployed have until December 2, 2026, to comply.
This regulatory obligation addresses a specific problem: without labeling, a reader cannot distinguish an article written by a journalist from a text assembled by a language model. The labeling of AI-generated content acts as a transparency signal comparable to the mention of an author or source in traditional media.
The measure does not solve everything. A machine-readable technical label is only useful if dissemination platforms display it in a way that is understandable to the end user. The regulation sets the framework, but the trust chain also depends on the interface choices of search engines, social networks, and aggregators.
Resources dedicated to the circulation of information document these regulatory developments and their practical consequences, such as https://www.informationinflux.org/, which gathers analyses on these topics.
AI assistants and generative engines: new intermediaries of information
Traditional search engines directed users to source pages. Assistants and generative engines produce a synthesized response directly, without always redirecting to the original content. This change profoundly alters the relationship with the reliability of online information.
When an AI assistant summarizes a news article to answer a question, the reader receives a response without editorial context. They see neither the author’s name, nor the media’s editorial line, nor the publication date. These elements, which usually help assess the credibility of a source, disappear from the interface.
Consequences for media funding
Data from ACPM reported in September 2026 indicate a marked decline in visits to several major news sites in August 2026. Publishers are questioning the effect of automated summaries provided by Google. If the interface provides the answer without sending the user to the source, the media’s business model becomes fragile.
A media outlet that loses its direct traffic also loses its advertising revenue and subscriptions. In the long term, financing independent and verified information production becomes more difficult. The issue goes beyond the technological framework: it touches on the public’s right to access information produced according to journalistic standards.
Media education and source verification: reflexes to instill
In the face of the multiplication of channels and formats, the individual capacity to evaluate a source becomes a fundamental skill. Several criteria allow for assessing the reliability of digital content:
- The identification of the author or producing organization, with a clear mention of their qualification or editorial mandate
- The presence of cited primary sources (official documents, raw data, attributed testimonies) rather than unsupported claims
- The publication and update date, which allows for verifying that the information remains valid in its current context
- Cross-referencing with at least one other independent source covering the same topic
These reflexes apply to both news articles and responses provided by AI assistants. Cross-referencing between independent sources remains the most reliable verification mechanism, regardless of the dissemination medium.
The role of libraries and public institutions
Historically, libraries have played a role in sorting and providing organized information. In the digital world, this role evolves towards supporting information literacy. Training users to distinguish a reliable source from unverified content is part of the missions that many libraries integrate into their programs.
Public institutions also contribute to this effort. UNESCO reminds us that 139 countries today have laws on freedom of information, covering the vast majority of the global population. This legal framework guarantees a right to access public information, but its effectiveness depends on citizens’ ability to seize it.
Misinformation and social networks: the mechanisms of propagation
Social networks amplify the dissemination of unverified content by their very architecture. Recommendation algorithms favor posts that generate engagement, regardless of their accuracy. A false but emotionally charged piece of content often spreads faster than a nuanced factual analysis.
Misinformation exploits the speed of social media dissemination and the lack of friction in republishing. Sharing content takes just a gesture, while verifying its accuracy requires several minutes of research. This structural imbalance favors misinformation.
The AI Act and the labeling obligations for AI-generated content represent a first regulatory response. Other complementary approaches exist: fact-checking mechanisms integrated into platforms, trust labels awarded to media that meet transparency criteria, or initiatives for source certification by independent organizations.
Access to reliable information does not depend on a single lever. It relies on a combination of regulation, technological transparency, and individual skills. The European framework being established with the AI Act sets traceability obligations, but the final verification remains that of the reader who cross-references their sources before granting their trust.
