The European Parliament, the institution that wrote the world’s first comprehensive AI law, is building its own internal AI platform after discovering that roughly 20 percent of its staff were using public chatbots, including ChatGPT, to draft legislation.
The platform, called EPGenAI Hub, is currently in testing and could begin rolling out as early as September, according to Politico. It will give MEPs and staff access to models from Meta, OpenAI, Anthropic, and France’s Mistral through a controlled environment designed to keep parliamentary data off unvetted public services.
The problem: AI slop in legislation
The scale of unsanctioned AI use inside the Parliament came into focus at a meeting in early July, when several committee chairs raised alarms about the growing volume of AI-generated amendments, speeches, and parliamentary questions. The signs were unmistakable: a surge in amendments needing translation, recognizable AI drafting patterns, and citations to laws that do not exist.
“Parliament has not been immune to AI slop and misleading AI-generated content,” said Brando Benifei, the Socialist MEP who steered the AI Act through the Parliament. “MEPs must lead by example as the AI Act’s transparency rules become applicable.”
The AI Act’s transparency obligations, including Article 50 requirements for disclosing AI-generated content in matters of public interest, become enforceable on August 2, 2026. Yet MEPs themselves will not be required to disclose when they use EPGenAI Hub to draft the laws governing everyone else, a contradiction that critics have already flagged.
Two tiers of access
EPGenAI Hub is designed around a sensitivity-tiered architecture. For routine tasks, users can choose locally operated models running inside Parliament’s own infrastructure, including Meta’s open-weight Llama, OpenAI’s GPT-OSS, and Mistral Small. These keep sensitive material on the institution’s systems. For more demanding work, users can access external services including ChatGPT and Claude Sonnet, trading containment for capability.
The platform has already been demonstrated to MEPs and staff at workshops pitched around working “smarter and faster.” The Parliament estimates that roughly 2,100 people, about one in five staff members, use AI tools daily, and much of that usage currently runs through uncontrolled public platforms.
A European irony
The choice of providers carries its own political dimension. The Parliament that wrote the AI Act and champions European technological sovereignty is building its platform primarily around US models from Meta, OpenAI, and Anthropic, alongside one French provider. European open-source alternatives remain less competitive at the frontier, and Parliament’s leadership judged that restricting access to European-only models would simply drive staff back to unauthorized public tools.
EPGenAI Hub is expected to launch with a soft rollout in September, followed by broader access later in the year. The Parliament has not disclosed the platform’s total cost or whether it will eventually be made available to national parliaments across the EU.
Sources: The European Parliament’s answer to its AI worries: More AI (Politico, July 2026); European Parliament builds its own AI platform for MEPs (TNW, July 2026); European Parliament creates its own AI platform (Cybernews, July 2026)