The European Commission continues to develop advanced AI-based multilingual services to facilitate communication across languages. These services – developed through the Digital Europe programme – support a more inclusive and competitive Europe, providing hands-on tools to users including public administrations, civil society and small businesses.
Free, secure AI-based language tools to translate, generate and improve content in multiple languages: eTranslation, eBriefing, eReply, eSummary, Anonymisation, WebText, Accessible Text, Multilingual Post and Speech-to-Text.
A history of innovation – from eTranslation to an evolving package of AI-based multilingual services
The European Commission has been involved in machine translation since the 1980s, expanding into an internally built system in 2013. This was based on the data available in Euramis (European Advanced Multilingual Information System) – a vast repository of high-quality multilingual data from the translation services of all the EU institutions.
In 2015, the European Language Resource Coordination (ELRC) initiative began gathering language data across all 24 of the EU’s official languages. Crucially, this initiative
- collected valuable data both for large languages (e.g. English, French) and for under-resourced languages (e.g. Maltese, Danish, Slovenian)
- boosted the training potential for smaller languages, combating the risk of digital extinction.
The data collected was instrumental in training the neural machine translation engines that power eTranslation today.
Building on advances in computing power and AI, and the wider language coverage offered by the ELRC initiative, the Commission launched eTranslation in 2017. This marked a significant step forward, as neural technology mimics the human brain, allowing more accurate and nuanced translations.
Since then, eTranslation has developed into a world-class machine translation service operating within a highly secure environment. In its first year it delivered 19 million pages, and by 2024 production had grown to 764 million pages.
Thanks to the Commission’s experience with big data and the success of eTranslation, it has been able to build up the skills needed to develop other AI-based multilingual services.
- 2025
Web Text, Accessible Text
- 2024
eBriefing, eReply
- 2023
eSummary
- 2022
Anonymisation support
- 2021
Speech-to-Text
What DG Translation does
One of the pillars of DG Translation’s core business of translating for the European Commission is eTranslation and other language technology and AI-based services. Our AI-based services team works with the Directorate-General for Communications Networks, Content and Technology (DG CONNECT) to develop these tools for the Commission, other EU institutions and eligible users across the EU and countries affiliated with the Digital Europe programme.
They lead digital transformation initiatives in DG Translation and encourage innovation and novel ideas by supporting user-driven technology watch activities, particularly in the field of AI. Through the Commission’s internal AI network, they play a key role in exploring AI-based technologies within the Commission and beyond.
This multidisciplinary team of engineers, developers, linguists, computational linguists, prompt engineers and communication specialists works consistently to develop and promote new AI-based multilingual services.
EU language data – what is Euramis?
Euramis (European Advanced Multilingual Information System) is a system of databases or memories – and associated tools – fed by the various EU institutions. It stores the history of documents drafted and handled by the EU institutions, along with their translations.
These may be in several or all EU official languages, and are cross-referenced so that they are easy to reuse for incoming translation requests. Documents are broken down into segments – full sentences, phrases or single words. These are all labelled by language.
The linguistic services of the EU institutions have a long history of handling data. One of the key challenges of public service translation is the need for consistency across large volumes of text. Euramis started in 1995 as a reference database for Commission translators and has developed into a vast repository of high-quality data from the translation services of all EU institutions. It now contains over 100 billion tokens (roughly 270 million pages).
Hundreds of thousands of segments are translated by EU translators every day. This high-quality data is checked, filtered, quality controlled and curated by professional linguists in all 24 official EU languages, and stored in the Euramis database – the basis for eTranslation. The European Commission is also using multilingual data from Euramis to develop an EU institutional LLM for future use by EU public administrations, small businesses, academia and non-governmental organisations.
An EU large language model to promote language equality
Although Euramis is not directly accessible outside the EU institutions, since 2007 DG Translation has made its multilingual Translation Memory for the Acquis Communautaire (DGT-TM) available to the public.
This is one way we contribute to the European Commission’s general effort to support multilingualism, language diversity and the reuse of Commission information.