Artificial intelligence is already part of the daily life of many newsrooms. It is used to generate drafts, summarize information, translate content, transcriptions, create images, or support different editorial tasks.
As of August 2, 2026, digital media will have to pay more attention to how they inform readers when content has been generated, modified, or assisted by artificial intelligence.
The change does not imply that any use of AI must be automatically labeled. The rule introduces a broad, but not absolute, transparency obligation. The key point is to assess whether the intervention of AI in the content is substantial and if it can lead the reader to confusion about the origin, authorship, or authenticity of what they are consuming.
This means the line is not always clear. It is not the same to use an AI tool to correct a text, summarize information, propose a headline, or make a supervised translation, as it is to publish a news story generated entirely by AI, create an artificial image, or significantly modify audiovisual content.
In the informational sphere, the same regulation takes into account human supervision and editorial responsibility. When content created with AI assistance is reviewed, edited, and validated by a newsroom before publication, the obligation to inform the reader may not be necessary in all cases. But this decision can no longer be left to improvisation: each medium should have its own clear and documented criteria.
Therefore, the challenge for editors is not just deciding whether or not to display an "AI-assisted content" label. The challenge is to have an internal process that allows knowing when AI has been used, with what intensity, with what impact on the final content, and if it is necessary to communicate it to the reader.
The regulation leaves room for interpretation and obliges media to make editorial decisions. Precisely for this reason, at Comitium we recommend preparing before these obligations come into force and defining a common policy for the entire newsroom.
What do we recommend to editors?
The best answer is not to label everything by default nor to never label. The best answer is to establish a coherent editorial criterion.
At Comitium, we recommend that each medium define, at a minimum, three levels of AI use:
-
Simple editorial assistance
Corrections, headline proposals, internal summaries, supervised translations, or help with text structure. In these cases, if there is human review and editorial responsibility, we do not see it necessary to inform the reader. -
Relevant assistance in published content
Texts partially generated with AI, significant rewrites, automatically published summaries, highlights, visible tags, or content where AI has had a real impact on the final result. In these cases, the media outlet should consider displaying a tag such as “AI-assisted and editorially supervised content”. -
Content substantially generated or manipulated with AI
Entirely generated texts, synthetic images, modified videos, or any content that could mislead about its origin or authenticity. In these cases, the information to the reader should be clear and visible.
This criterion should be accompanied by three practical measures: registering the use of AI within the CMS, always guaranteeing human review before publishing, and training the newsroom to know when content needs to be marked as AI-assisted.
What if you use external tools like ChatGPT or Gemini?
Many newsrooms combine different artificial intelligence tools, such as ChatGPT, Gemini, Claude, Copilot, or other specialized platforms.
In these cases, it is advisable for the organization to have an internal procedure to identify when these tools have participated in the creation of content and to define which cases require communication to the reader.
Having this information also facilitates the traceability of the editorial process and allows demonstrating that there has always been human supervision.
What does Comitium offer?
At Comitium, we have long integrated artificial intelligence as a support tool for editorial teams, always keeping control in the hands of the newsroom.
Clients who use Comitium's AI functionalities already have a specific field to indicate whether content has been generated or assisted by artificial intelligence, which facilitates the registration of this information within the editorial workflow.
This same criterion can also be applied to content created with external tools, allowing each media outlet to maintain consistent traceability regardless of the technology used.
Looking to the future
The deployment of the AI Act will continue over the coming months and years. It is foreseeable that new standards, recommendations, and best practices related to the identification of AI-generated content will emerge.
For this reason, at Comitium, we will continue to evolve the platform so that our clients have the necessary tools to adapt to the new European framework with maximum ease.
Artificial intelligence represents a great opportunity for digital media, but it also demands transparency, responsibility, and trust. Having a well-defined editorial process will be the best way to leverage all its benefits while maintaining credibility with readers.
More information
If you wish to delve deeper into the transparency obligations introduced by the European Artificial Intelligence Regulation (AI Act) and its impact on AI-generated content, we recommend reading this article from PwC: The European Artificial Intelligence Regulation will require AI-generated content to be explicitly labeled.
