Microsoft unveiled two major initiatives in Paris designed to preserve Europe"s linguistic and cultural heritage and bolster the continent"s position in the AI era. Building on its earlier European Digital Commitments, to expand AI and cloud infrastructure, strengthen data privacy, enhance cyber‑resilience, and support Europe’s broader digital competitiveness, these new steps will make European languages and cultural assets more accessible online and better represented in large language models (LLMs).
Europe boasts over 200 languages and a millennia‑old cultural legacy that underpins both creative expression and economic activity by facilitating communication, innovation, and trade. However, as online content becomes increasingly dominated by English, much of it reflecting an American perspective, Europe’s commerce and culture risk being underserved in the data that trains today’s large language models. Microsoft"s Vice Chair and President, Brad Smith, states, "AI that doesn’t understand Europe’s languages, histories, and values can’t fully serve its people, its businesses, or its future."
To illustrate this imbalance, Llama 3.1, an open‑source model, scores over 15 points lower in Greek and more than 25 points lower in Latvian than in English, placing it top of the class in English, mid‑range in Greek and near the bottom in Latvian, a pattern seen across major LLM benchmarks.
To address this, Microsoft will base expertise at its innovation centres in Strasbourg, France, to develop and curate multilingual datasets on Microsoft Azure. These teams from its Open Innovation Centre (MOIC) and AI for Good Lab will work with cultural institutions, academic partners, and technology firms across Europe to expand the availability of training data in ten under‑represented European languages, among them Estonian, Alsatian, Slovak, Greek, and Maltese.
Microsoft has also issued a call for proposals to source digital texts, transcripts, and other materials suitable for AI development. Applications for grants, which will provide recipients with Azure credits and engineering and technical support, will be available on the AI for Good Lab website beginning on September 1st, 2025.
Secondly, Microsoft is extending its Culture AI programme this autumn to create a high‑fidelity digital replica of Notre Dame Cathedral in Paris. In partnership with the French Ministry of Culture and the heritage‑digitisation specialist Iconem, the project will capture the 862‑year‑old Gothic landmark in fine detail. Previous Culture AI projects have digitally preserved sites such as Ancient Olympia in Greece, Mount Saint‑Michel in France, St Peter’s Basilica in Rome and the Normandy Allied beach landing sites.
The tech giant says that these initiatives build on Microsoft’s more than 40 years of localisation experience. Today, Windows supports over 90 languages, including all official European Union languages as well as regional tongues such as Basque, Catalan, Galician, Luxembourgish and Valencian. Microsoft 365 also provides Office interfaces in more than 30 European languages. By integrating Europe’s languages and cultural assets into its AI and cloud platforms, Microsoft aims both to safeguard the continent’s heritage and to empower its businesses and citizens in the digital era.
The company emphasises that these steps are purely supportive, contributing open data, tools, and expertise rather than proprietary assets.