Breaking the language barrier: How De Standaard is testing a French edition through AI translation
Flemish news brand De Standaard, part of the Mediahuis group, has launched De Standaard en français, a digital edition translating selected articles from its native Dutch newspaper to French. By experimenting with AI-supported translation workflows, Mediahuis aims to reach a new, French-speaking subscription market without building a separate newsroom.
A bold experiment in a divided media landscape
Belgium’s media landscape is divided along linguistic lines. Flemish outlets mainly address Dutch-speaking audiences, while French-speaking readers are served by different news organisations. As a result, there are very few national media brands that reach audiences across the entire country.
Editors at De Standaard began discussing how new technologies could help overcome this barrier. The idea emerged during an informal conversation about AI and its potential role in journalism. Could AI make it possible to translate the newspaper into another language and enter a new audience market?
From that conversation, the concept of De Standaard en français was born. The initiative was designed as a lean experiment that could test a new audience segment without fundamentally changing the newsroom structure.
A digital edition for French-speaking readers
The French edition is published daily as a digital product. Each morning from Monday to Saturday, readers receive a curated selection of 20 to 25 articles translated from the Dutch edition of De Standaard. The selection focuses on Belgian and international politics, economics, investigative journalism, opinion pieces and cultural coverage.
The product is available through a web app and newsletter, which allows the team to launch quickly without developing a separate mobile application. The target audience mainly consists of French-speaking readers in Brussels and larger cities in Wallonia, particularly those interested in in-depth journalism about Belgian and European affairs.
When the product launched in December 2025, readers were given access through a free trial period running until the end of March. The long-term goal is to convert this audience into a paying subscription base.
Integrating the project into the newsroom
To keep the experiment manageable, the team reused existing newsroom structures wherever possible. The same news managers who decide which stories appear in the Dutch edition, also determine which articles are selected for translation. Instead of creating original French content, the project translates journalism that already exists within the newsroom.
A small bilingual team oversees the translation process and editorial review. The first version of each translation is generated with ChatGPT. They use an extensive prompt that specifies the tone of voice, editorial style and translation guidelines. However, every article still passes through a human editorial check before publication to ensure the text remains natural and faithful to the original journalism.
Key takeaways for media organisations
- AI translation enables efficient expansion to audiences in other languages. By building on existing editorial workflows and infrastructure, media organisations can launch new language products without creating separate editorial teams or adding major operational complexity.
- Human oversight remains essential. While tools like ChatGPT can generate the first version of a translation, editors are still needed to safeguard journalistic quality and ensure the brand’s tone of voice remains consistent.
- Lean experimentation lowers the barrier to testing new audiences. AI translation allows publishers to experiment with new language products before committing significant resources, while also making journalism accessible to audiences who were previously excluded due to language barriers.
By leveraging existing content and workflows, media organisations can explore new markets and business models in a fast and low-risk way, before committing to larger investments.
