The Dutch Data Protection Authority (AP) has urged voters not to rely on AI chatbots for election advice, warning that the systems deliver unreliable and biased recommendations ahead of the October 29 national election. The regulator found that chatbots frequently directed users toward just two major political parties — the far-right Freedom Party (PVV) and the Labour-Green Left coalition — despite the Netherlands’ highly fragmented political landscape.
According to the AP’s tests, chatbots advised users to vote for one of those two blocs in 56% of cases, even when provided with the campaign programs of smaller parties. “Chatbots may seem like clever tools, but as a voting aid, they consistently fail,” said Monique Verdier, the watchdog’s vice-chair, adding that their internal operations are “unclear and difficult to verify.”
While the watchdog did not identify the four chatbots tested, it warned that their underlying algorithms may be inadvertently promoting political polarization by amplifying dominant parties on opposite ends of the spectrum. Current polls predict around 20% support for the Freedom Party and 16% for the Labour-Green Left coalition, highlighting their strong influence in public discourse.
The Dutch election follows the collapse of a right-wing coalition earlier this year, leaving the country under a caretaker government and setting the stage for a contest between conservative and centrist forces. Although it is unclear how many citizens are turning to AI tools for political guidance, the regulator said their use is “growing,” noting that more than 13 million voters are eligible to participate.