European Parliament Develops Dedicated AI Platform for MEPs
The European Parliament is addressing a unique challenge: MEPs (Members of the European Parliament) and their staff are utilizing AI tools to draft speeches, questions, and amendments, raising concerns about potential impacts on legislation.
July 20, 2026 – 12:42 pm
(Image by Canva)
The Solution: The Parliament is introducing EPGenAI Hub, an internal platform providing access to AI models from Meta, OpenAI, Anthropic, and France’s Mistral. Testing is underway, with a potential rollout as early as September, according to Politico.
Background: AI has integrated into the daily work of the Parliament, with approximately 2,100 staff members (around 20%) using it regularly. This raises questions about its influence on legislative processes. At a recent meeting, committee chairs expressed concerns over recognizable drafting styles, translation demands, and even hallucinated references in AI-generated content.
Brando Benifei, the lawmaker who steered the EU’s AI Act, emphasized the need for MEPs to set an example regarding transparency rules.
Features of EPGenAI Hub:
- Users can select models based on task sensitivity.
- Some models run within the Parliament’s systems (e.g., Meta’s Llama, OpenAI’s GPT-OSS, Mistral Small), while others are powerful external services (ChatGPT, Anthropic’s Claude Sonnet).
- The primary goal is to prevent MEPs from using uncontrolled public AI apps for parliamentary material.
Unaddressed Governance Issues:
Currently, there are no obligations for MEPs to disclose whether an amendment, question, or speech was AI-assisted. Committee chairs requested the secretary-general to develop guidelines, which may fall to the constitutional affairs committee. Staff already have rules labeling AI-assisted work and prohibiting sensitive information in public AI tools, due for a rewrite.
Divergent Perspectives:
Not all MEPs are comfortable with this shift. Barry Andrews, an Irish Renew MEP, acknowledges the workload but expresses his personal belief that core democratic functions should not be entrusted to AI.
Contradiction: The rollout relies heavily on American AI models despite Brussels’ efforts to promote digital sovereignty and reduce dependence on US technology.