European Parliament launches EPGenAI Hub to control AI use by lawmakers

The European Parliament is testing a new internal platform, EPGenAI Hub, that will give members and staff access to generative‑AI models from OpenAI, Meta, Anthropic and France’s Mistral. The hub is intended to replace use of public AI tools, allowing the institution to enforce safety and transparency guidelines as the AI Act’s rules become applicable.

Around 2,100 people – roughly 20% of Parliament staff – already use AI daily for drafting speeches, amendments and translations. Officials have raised concerns about AI‑generated errors, hallucinated references to non‑existent laws and the overall quality of legislative work. The hub, still in testing, could be rolled out as early as September and will require AI‑assisted content to be clearly labelled.

The initiative also follows security steps such as replacing Google services with the EU‑based Qwant search engine and disabling AI functions on parliamentary tablets, reflecting broader efforts to limit dependence on unsecured external AI services.