European Parliament launches internal AI platform for MEPs
▼ Summary
– The European Parliament is developing an internal AI platform, EPGenAI Hub, to control the widespread, unauthorized use of public chatbots by MEPs and staff for legislative work.
– About 20% of Parliament staff use AI daily, causing issues like a surge in amendments needing translation, recognizable AI drafting styles, and hallucinated legal references.
– The Hub offers a choice of models, including Meta’s Llama and OpenAI’s GPT-OSS, running internally or externally, to prevent sensitive parliamentary material from being fed into uncontrolled public apps.
– There are no rules requiring MEPs to disclose AI use in amendments or speeches, though staff must label AI-assisted work under existing guidelines that are being updated.
– The initiative creates a contradiction as it relies heavily on US AI firms while Brussels pushes for digital sovereignty, though open-source and EU models are under review.
The European Parliament faces a growing problem with artificial intelligence, and its solution involves deploying even more of it. Lawmakers have increasingly turned to public chatbots to assist in drafting legislation, prompting the institution to develop an official platform designed to bring this practice under institutional control.
The new internal system, known as the EPGenAI Hub, will provide Members of the European Parliament (MEPs) and staff with access to models from Meta, OpenAI, Anthropic, and France’s Mistral. Currently in a testing phase, the platform could see its rollout begin as early as September, according to Politico.
AI has quietly integrated into the Parliament’s daily operations. Approximately 2,100 people, or about 20% of staff, now use it every day. The concern is what this reliance means for the legislative process itself.
During a meeting this month, several committee chairs raised alarms that MEPs and their assistants were depending on AI to draft amendments and other legislative documents. The signs are unmistakable: a noticeable increase in amendments requiring translation, recognizable drafting patterns, and occasional hallucinated references to laws that do not exist.
“Parliament has not been immune to AI slop and misleading AI-generated content,” said Brando Benifei, the lawmaker who guided the EU’s AI Act through negotiations. “MEPs must lead by example as the AI Act’s transparency rules become applicable.”
The EPGenAI Hub allows users to select a model based on the sensitivity of their task. Some models, including Meta’s Llama, OpenAI’s GPT-OSS, and Mistral Small, run entirely within the Parliament’s own systems. Others, like ChatGPT and Anthropic’s Claude Sonnet, are more powerful services hosted externally.
The main objective is to prevent officials from feeding parliamentary material into uncontrolled public applications. The Parliament has already demonstrated the platform to staff during workshops, one of which promised to help them “work smarter and faster.”
However, the rules have not kept pace with the technology. MEPs face no obligation to disclose whether an amendment, question, or speech was generated using AI. Committee chairs have asked the secretary-general to draft guidelines, but they were told the task may instead fall to the constitutional affairs committee, which would need to rewrite the Parliament’s internal rules.
Staff members already operate under their own rules, established in April 2024 and now due for revision. These require labeling AI-assisted work and prohibit inputting sensitive information into public AI tools.
Not everyone is comfortable with this direction. Barry Andrews, an Irish Renew MEP, noted that some colleagues clearly use AI to draft questions, speeches, and amendments. “I fully understand, given our workload,” he said. “But for me personally, these are core democratic functions which voters elected us to perform.”
There is an awkward contradiction at play. The rollout relies heavily on American AI, even as Brussels pushes to reduce its dependence on US technology. The Parliament has pursued digital sovereignty elsewhere, swapping Google for the French search engine Qwant and disabling built-in AI on MEPs’ tablets over data-protection concerns. Now it is opening its doors wider to US firms.
This mix may not last. The Parliament’s press service said the models currently on offer are under review. Open-source and EU sovereign models are becoming available through cloud providers recently contracted by the European Commission. For now, the institution writing Europe’s AI rulebook is still figuring out how to use the technology itself.
(Source: The Next Web)
