EU Lawmakers Face Risks From AI Overreliance
Parliament is drafting new guidelines after AI-written amendments cited laws that do not exist, a research-stage governance response.
Updated on Nov. 4, 2026 in Artificial Intelligence
Live Poll
Do you think lawmakers using AI to draft laws can still be held accountable?
European Parliament President Roberta Metsola warned lawmakers on Wednesday not to become overreliant on artificial intelligence in decision-making. The warning is research-stage in governance terms: Parliament is preparing new AI guidelines for lawmakers, which its administration will submit to senior leaders very soon.
Why it matters
Parliament officials have warned of a surge in AI-written amendments, some of which reference laws that do not exist, showing how generative AI tools can introduce errors into legislative work. The warning matters for the trajectory of AI in public institutions, where the question is shifting from whether to use commercial models to how to contain their failure modes.
The evidence is qualitative: Parliament officials reported a surge in AI-written amendments, and some cited laws that do not exist. Parliament has not published counts of affected amendments, and the detailed error rates of the tools in use were not disclosed.
The players
European Parliament
The EU legislature, which is preparing new AI guidelines for lawmakers and is building an internal AI tool hosted on its own infrastructure.
Roberta Metsola
President of the European Parliament, who warned MEPs against overreliance on AI in decision-making.
The details
Parliament is building its own AI tool for sensitive work, hosted on its internal infrastructure, so that confidential material stays within its own systems. Commercial models such as ChatGPT, Claude and Meta's Llama can be used for less sensitive tasks, but they generate text by predicting likely word sequences, which can produce confident citations that do not correspond to real legislation.
Timeline
2024: Parliament issued its existing guidelines for staff on AI use.
July: POLITICO reported on the use of commercial AI models in Parliament.
Wednesday: Metsola delivered the warning to MEPs at an event on AI use.
The Tech Race
The episode places Parliament among institutions adopting commercial generative models while building in-house alternatives for sensitive work. Its effort is a governance response within the broader race to deploy AI in government, where internal tooling is becoming the standard for confidential material.
Parliament's new guidelines will govern how staff use AI on legislative work, though their publication date has not been announced. Readers should expect tighter rules on commercial models for sensitive tasks, while the timeline for the in-house tool's rollout has not been specified.
The takeaway
The central lesson is that commercial AI can produce plausible but nonexistent legal references, so verification remains a human responsibility. Watch for the publication of Parliament's new AI guidelines, which its administration will submit to senior leaders very soon.
Further reading
For broader context on how these tools are being adopted across institutions, see Artificial Intelligence.
Live Poll
Do you think lawmakers using AI to draft laws can still be held accountable?
