Metsola urges MEPs to retain responsibility as Parliament prepares AI guidance
Parliament is developing an internal AI tool for sensitive tasks, while officials warn that AI-generated amendments have cited non-existent laws.
European Parliament President Roberta Metsola urged lawmakers on 4 November 2026 to keep human judgement central to their work as the institution prepares guidance on artificial intelligence. Speaking at a Parliament event, she said AI assistance with amendments and legal research could not substitute for political discussion or critical thinking.
Metsola called for transparency about AI use and said elected representatives remained accountable for decisions regardless of how they employed the technology.
Secretary-General Alessandro Chiocchetti said Parliament was developing an AI tool on its own infrastructure to handle sensitive work. Commercial models, including ChatGPT, Claude and Meta’s Llama, remain available for tasks involving less sensitive material.
Parliament officials have warned that growing numbers of AI-generated amendments include references to laws that do not exist. Chiocchetti described a case in which an MEP received incorrect legal advice from AI, saying Parliament’s lawyers needed two days to demonstrate that ChatGPT’s answer was wrong.
Chiocchetti said the administration would shortly put new AI guidelines before senior leaders. Staff guidance currently in place dates from 2024. He outlined three principles: disclose AI use as required, keep sensitive information out of public tools and verify material before sending it out.
He also said he wanted AI to help organise thousands of amendments, while insisting that MEPs retain control over drafting and scrutinising legislation. People must write, check and take responsibility for texts put to a vote, he said.
