AI chatbots are "unreliable and clearly biased" when offering voting advice, the Dutch Data Protection Authority (AP) has said, warning of a threat to democracy eight days before national elections.
The four chatbots tested by the AP tended to advise people to vote for parties on the left or on the far-right extreme of the political spectrum, the watchdog tested.
The machines "often end up with the same two parties, regardless of the user's question or command", the AP said in a new report ahead of the 29 October election.
In more than half of the cases, the chatbot suggested either the far-right Freedom Party (PVV) of Geert Wilders or the left-wing GroenLinks-PvdA led by former European Commission vice president Frans Timmermans.
Some parties, such as the centre-right CDA, "are almost never mentioned, even when the user's input exactly matches the positions of one of these parties".
The deputy head of the AP, Monique Verdier, said that while chatbots might seem like clever tools, "as a voting aid, they consistently fail".
Voters are being pushed towards a party that does not necessarily align with their political views, she added.
"This directly impacts a cornerstone of democracy: the integrity of free and fair elections," added Ms Verdier.
"We therefore warn against using AI chatbots for voting advice, as their operation is unclear and difficult to verify."
The Dutch head to the polls on 29 October in an election that will be closely watched around Europe for the performance of the far-right PVV.
The PVV of Wilders is currently leading the polls but the gap to the GroenLinks-PvdA and CDA appears to be narrowing, with many Dutch yet to make up their minds.
All major parties have ruled out an alliance with the PVV, meaning the party that comes second is most likely to provide the next prime minister.
The AP emphasised the bots were not deliberately biased, their "identified shortcomings are a consequence of the way AI chatbots operate".
