 "Election campaign boards are displayed ahead of the Dutch parliamentary election on October 29, in The Hague, Netherlands, on October 7, 2025. Reuters File")
The Dutch data protection authority has urged voters not to rely on artificial intelligence chatbots for voting advice ahead of the country’s general election next week, citing concerns over transparency and bias.
“AI chatbots give a highly distorted and polarised image of the Dutch political landscape in a test,” the data protection watchdog warned in a study published on Tuesday.
“We warn not to use AI chatbots for voting advice, because their operations are not transparent and verifiable,” Politico quoted Monique Verdier, vice-chair of the authority, as saying in a statement.
She called upon the chatbot developers to “prevent that their systems are being used for voting advice.”
Dutch voters elect a new parliament next Wednesday.
The Dutch data protection authority has raised concerns over the role of AI chatbots in political guidance, after an experiment revealed skewed and polarised voting advice ahead of next week’s general election.
In a test conducted across four major AI chatbots — OpenAI’s ChatGPT, Google’s Gemini, Elon Musk’s Grok, and Mistral AI’s Le Chat — the watchdog found that voter profiles were often directed toward political extremes, with centrist parties largely underrepresented, reported Politico.
The authority created a series of voter profiles based on validated Dutch voting-aid tools and asked each chatbot to provide voting advice tailored to those profiles.
Voters on the progressive left were “mostly directed to the GreenLeft–Labor” party led by former European Commission executive Frans Timmermans, while those with right-leaning views were “mostly directed to the PVV,” the far-right party led by Geert Wilders, which is currently leading in the polls.
Impact Shorts
More ShortsDespite being evenly represented in the constructed profiles, centrist parties received little to no mention in the responses generated by the chatbots, added the report.
All four companies have made partial or full commitments under the EU’s voluntary code of practice for advanced AI models, which calls on developers to mitigate societal and rights-based risks posed by their systems.
OpenAI, Google, and Mistral are full signatories, while Grok’s parent company xAI has signed up to parts of the code.
The Dutch authority also suggested that chatbots offering voting advice could fall under the “high-risk” category of the EU’s upcoming AI Act, which would subject them to stricter regulations starting mid-2025.
With inputs from agencies