European Parliament: In-House AI to Support Its MEPs

The Rise of AI in the European Parliament

Artificial intelligence has become a central topic within European institutions, generating both interest and concern. The European Parliament has taken concrete steps to regulate the use of this technology among its members.

The European Commission recently launched an ambitious project aimed at developing an open-source artificial intelligence model capable of processing the 24 official languages of the European Union. This initiative is part of a broader strategy where the EU seeks to position itself against global AI giants such as China and the United States, while also regulating its use within its own structures.

Growing Concerns in Parliament

The use of artificial intelligence by Members of the European Parliament has become commonplace, particularly for drafting amendments and finalizing legislative work. According to reports from Politico, this trend sparked debates during a meeting last July, where committee chairs expressed their concerns.

Brando Benifei, a Member of the European Parliament from the Progressive Alliance of Socialists and Democrats, emphasized that the Parliament is not immune to errors and misleading content generated by AI. Authorities are also worried about the quality of work produced with AI assistance and the sharing of parliamentary documents with private platforms not controlled by Brussels. This situation has prompted authorities to consider solutions to regulate this growing usage.

EPGenAI Hub: An Institutional Response

In response to these concerns, the European Parliament has established a new internal platform, EPGenAI Hub, to secure the use of AI. Currently, around 2,100 people, or one-fifth of the Parliament's workforce, use generative AI on a daily basis.

The EPGenAI Hub platform allows Members of Parliament to access local AI models, such as GPT-OSS, Mistral Small, or Llama, within infrastructures secured by the Parliament. Additionally, it also offers the possibility of using externally hosted models, such as ChatGPT or Claude Sonnet, which introduces American solutions along with the associated risks.

An Evolving Offering

The European Parliament is aware of the potential criticisms related to the use of these technologies. The Parliament's press service clarified that the offering of AI models is currently under review. This revision takes into account the availability of open-source models and sovereign models developed by sovereign cloud providers, which have recently signed a contract with the European Commission.

This approach aims to strengthen the digital sovereignty of the European Union while ensuring a secure and controlled use of artificial intelligence within its institutions.