Microsoft Is Changing the Linguistic AI Game in Europe
Microsoft has announced a bold new plan to invest heavily in Europe to produce digital data supporting artificial intelligence (AI) models in more than 12 languages. The move aims to promote linguistic diversity in AI technologies, which have so far been largely dominated by English.
Brad Smith, the company’s president, stated in an interview with Agence France-Presse on Monday, July 21, 2025, that the continued effectiveness of AI models depends on the availability of linguistically diverse data. He pointed out that the near-total reliance on English could weaken the performance of these models in other languages.
Smith emphasized that AI becomes significantly less effective when used in languages that lack sufficient data, which automatically pushes users toward English. To bridge this gap, Microsoft plans to open new research branches in the French city of Strasbourg starting in September. The goal is to produce digital content in at least ten EU languages, including Estonian and Greek.
The initiative will include digitizing non-English books and recording hundreds of hours of audio in various European languages. Microsoft confirmed that all resulting data will be open-source and publicly available, with no proprietary claims by the company.
This effort comes amid growing European calls for greater “digital sovereignty,” in response to the dominance of American and Chinese tech firms in the AI sector. Through this initiative, Microsoft aims to position itself as a company aligned with European values on privacy and technological independence.
In the same vein, Microsoft also announced a new digital project to reconstruct a virtual version of Notre-Dame Cathedral in Paris. The digital replica is set to be delivered to the French government this fall, in collaboration with the French Heritage Institute and the company Iconem. Additionally, Microsoft has partnered with the National Library of France and the Museum of Decorative Arts to digitize portions of their cultural archives.
