In a November 4 account, To Vima reported that European Parliament President Roberta Metsola cautioned members against relying too heavily on AI to draft amendments or research the law. AI-generated drafting, she said, could not replace debate among colleagues, and AI-assisted legal research could not replace human judgment on difficult decisions.
What the proposed Parliament guidance would cover
European Parliament Secretary-General Alessandro Chiocchetti described three principles for guidance then being prepared: disclose AI use, keep sensitive material out of public AI tools, and check AI-assisted work before sending it. The guidance was expected to go to Parliament leadership.
The report said staff guidance dating to April 2024 required them to label AI-assisted work. At the time of the November 4 account, MEPs were not required to disclose AI use in amendments, parliamentary questions or speeches.
Officials reported errors in AI-assisted work
Parliament officials told To Vima they were concerned that some AI-written amendments cited laws that do not exist. Chiocchetti also recounted one case in which an MEP received incorrect AI-generated legal advice; Parliament lawyers reportedly took two days to disprove it.
The Commission’s legal analysis is a separate initiative
To Vima reported that the Parliament was developing an internally hosted AI tool for sensitive work. Separately, upday described European Commission systems for identifying similarities, possible overlaps, inconsistencies and duplication among EU legal acts, attributing its account of those systems to POLITICO. The Commission initiative concerns analysis of legal acts; the Parliament’s reported plans concern guidance for MEPs and a tool intended for sensitive institutional work.
