The European Parliament is building its own AI to stop MEPs pasting draft laws into ChatGPT

The European Parliament is building an internal AI platform called EPGenAI Hub to stop lawmakers from using public chatbots like ChatGPT to draft legislation, with a rollout as early as September. The tool will offer models from Meta, OpenAI, Anthropic and Mistral, but critics note the contradiction of relying on US AI while Brussels pushes for digital sovereignty. About 2,100 staff, or 20%, already use AI daily, and committee chairs have flagged a surge in AI-generated amendments with hallucinated references.

The European Parliament has a problem with AI, and its answer is more AI. Lawmakers keep using public chatbots to help write legislation. So the institution is building a sanctioned platform to bring that habit under control. The internal tool, called EPGenAI Hub, will give MEPs and staff access to models from Meta, OpenAI, Anthropic and France’s Mistral. The Parliament is testing it now and could begin the rollout as early as September, Politico reported https://www.politico.eu/article/the-european-parliaments-answer-to-its-ai-worries-more-ai/ . Why now AI has quietly seeped into the daily work of the Parliament. About 2,100 people, roughly 20% of staff, use it every day. The worry is what that is doing to the law. At a meeting this month, several committee chairs raised the alarm that MEPs and assistants were leaning on AI to draft amendments and other legislative work. The tells are hard to miss: a surge in amendments needing translation, recognisable drafting styles, and the odd hallucinated reference to a law that does not exist. “Parliament has not been immune to AI slop and misleading AI-generated content,” said Brando Benifei, the lawmaker who steered the EU’s AI Act https://thenextweb.com/news/oversight-board-ai-models-political-speech-censorship . “MEPs must lead by example as the AI Act’s transparency rules become applicable.” What the Hub offers The platform lets users pick their model based on how sensitive the task is. Some run inside the Parliament’s own systems, including Meta’s Llama, OpenAI’s GPT-OSS and Mistral Small. Others are more powerful services hosted outside, such as ChatGPT and Anthropic’s Claude Sonnet. The goal is to stop officials feeding parliamentary material into uncontrolled public apps. The Parliament has already shown it to staff at workshops, one of which promised to help them “work smarter and faster.” A governance gap The rules have not kept up. MEPs face no obligation to disclose whether an amendment, question or speech was written with AI. Committee chairs asked the secretary-general to draw up guidelines. They heard the job may instead fall to the constitutional affairs committee, which would have to rewrite the Parliament’s rules. Staff already have their own rules, from April 2024 and now due for a rewrite. They must label AI-assisted work, and cannot put sensitive information into public AI tools. Not everyone is comfortable with the drift. Barry Andrews, an Irish Renew MEP, said 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.” The sovereignty snag There is an awkward contradiction here. The rollout leans heavily on American AI, even as Brussels pushes to cut its dependence on US technology https://thenextweb.com/news/airbus-scaleway-aws-sovereign-cloud . The Parliament has been chasing digital sovereignty https://thenextweb.com/news/europe-ai-sovereignty-g7-vivatech elsewhere, swapping Google for the French search engine Qwant and switching off built-in AI on MEPs’ tablets over data-protection fears. Now it is opening its doors wider to US firms https://thenextweb.com/news/eu-google-android-ai-search-data-digital-markets-act . The mix may not last. The Parliament’s press service said the models on offer are under review. Open-source and EU sovereign models https://thenextweb.com/news/mistral-sovereignty-moment-anthropic-shutdown-le-chaton-fat are becoming available through cloud providers the European Commission recently contracted. For now, the institution writing Europe’s AI rulebook is still working out how to use the technology itself. Get the TNW newsletter Get the most important tech news in your inbox each week.