As MEPs and their staff increasingly rely on artificial intelligence to draft speeches, parliamentary questions, and even amendments, the European Parliament is moving to bring that activity under control, and its chosen solution is an officially sanctioned AI platform of its own.
The internal system, called the EPGenAI Hub, will give lawmakers and staff access to models from Meta, OpenAI, Anthropic, and France’s Mistral, offering what the institution hopes will be a safer alternative to the public tools already being used for legislative work. According to two people familiar with the plans, who spoke on condition of anonymity to discuss internal strategy, the platform is currently in testing and could begin rolling out as early as September.
The effort comes amid growing unease inside the institution over the spread of AI-generated parliamentary material. At a meeting earlier in July, several committee chairs raised concerns that a rising number of MEPs and assistants were using AI to draft amendments and assist with other legislative tasks, according to three people familiar with the discussion.
Brando Benifei, the Socialists and Democrats lawmaker who steered Parliament’s work on the EU’s landmark AI Act, said the institution had not escaped the wave of low-quality and misleading AI-generated content, and argued that MEPs should set an example as the AI Act’s transparency requirements take effect.
A dilemma at the heart of Parliament
The project reflects the broader challenge facing Parliament as generative AI becomes woven into its daily routines. Roughly 2,100 people, about 20 percent of all staff, now use AI on a daily basis.
Leadership wants officials to stop feeding parliamentary material into uncontrolled public platforms, but there is equal concern that the technology is eroding the quality of legislative work and introducing errors or misleading claims into the democratic process. Those suspicions have been reinforced by a surge in amendments requiring translation, recognizable drafting patterns, and the occasional invented reference to laws that do not exist.
Committee chairs asked Secretary-General Alessandro Chiocchetti to prepare guidelines for lawmakers’ use of AI. He suggested the issue might instead sit with the Parliament’s constitutional affairs committee, which would need to amend the institution’s rules of procedure to impose stricter requirements.
At present, MEPs face no obligation to disclose whether amendments, questions, or other work were produced with AI assistance.
Barry Andrews, the Irish Renew lawmaker who chairs Parliament’s development committee, said it was already evident that some MEPs were using AI to draft questions, speeches, and amendments. He said he understood the appeal given the workload, but added that for him personally these are core democratic functions that voters elected members to carry out.
Parliamentary employees, meanwhile, fall under separate internal rules on AI use. Those guidelines, dating from April 2024 and already due for revision, require staff to label AI-generated work with an “AI-assisted” disclaimer and bar them from entering sensitive information into public generative AI systems. The rules also caution that inaccuracies from the technology risk spreading misleading information or misrepresenting the Parliament.
Inside the new platform
The EPGenAI Hub has already been introduced to MEPs and staff at workshops, including one that promised to showcase practical everyday applications to help users work more efficiently, according to an email reviewed by reporters.
A beta version allows users to choose between models operated securely within Parliament’s own infrastructure and more powerful services hosted externally. The locally run options include Meta’s Llama, OpenAI’s GPT-OSS, and Mistral Small, while the externally hosted choices include OpenAI’s ChatGPT and Anthropic’s Claude Sonnet.
