As the Dutch parliamentary elections of October 29, 2025 approach, a warning has been issued by the Netherlands’ Data Protection Authority (AP). According to the institution, relying on artificial intelligence chatbots to guide your vote is a high-risk practice.
The findings from their recent tests highlight worrying biases in AI recommendations—an alarm bell that could echo far beyond Dutch borders, especially in France, where the 2026 and 2027 elections are fast approaching. Yiaho looked into the issue.
Biased AI in election advice?
The AP examined four artificial intelligence models to assess their ability to guide voters. The results are troubling: whatever the request, the chatbots tend to systematically favor two political parties—the Party for Freedom (PVV), a far-right party led by Geert Wilders, and the PvdA–GreenLeft alliance, led by Frans Timmermans.
Other parties, such as the Christian Democratic party CDA, are almost systematically ignored, even when users’ stated preferences match their platforms exactly.
This lack of neutrality raises questions about the reliability of AI in a context as sensitive as elections.
The problem lies in how these tools are designed: chatbots—from OpenAI to Yiaho to Gemini—rely on often opaque training data drawn from the internet, where information can be incomplete, outdated, or shaped by external dynamics.
As a result, the election advice they produce may skew voters’ perceptions by highlighting certain parties at the expense of others, without reflecting the diversity of the political landscape.
Read also on this topic: Anne Le Hénanff: Here is France’s new Minister for Artificial Intelligence
A growing and worrying trend?
This phenomenon is all the more concerning as more and more voters are turning to AI to inform their political choices.
In the Netherlands, this practice is gaining ground, and the AP fears that chatbot biases could distort the democratic process. This warning comes at a time when AI technologies are evolving at breakneck speed, making their potential influence on elections increasingly significant.
And in France, what impact will AI have in 2027?
If the Netherlands is sounding the alarm, France has every reason to pay attention. Looking ahead to 2026, the municipal elections will be a first test, before the 2027 presidential election, which is already shaping up to be a key moment.
AI, now everywhere in our daily lives, could play a decisive role in how election campaigns are run and perceived.
Candidates may be tempted to use these technologies to steer voters—for example by influencing algorithms or spreading optimized content to bias chatbot responses. Such manipulation, while technically complex, is not out of reach in a digital ecosystem where data is a formidable weapon.
And an AI president?
But an even bolder scenario could emerge! What if an AI itself ran as a candidate? Or joined a government team? In Albania, the government has already appointed the world’s first AI prime minister to oversee public procurement. An AI candidate, able to analyze voters’ expectations in real time and propose ultra-personalized solutions, could appeal to part of the electorate.
While this may sound like science fiction, rapid advances in artificial intelligence by 2027 could open the door to unprecedented debates about the place of machines in democracy. This scenario, though speculative, illustrates the scale of the upheaval AI could bring to politics.
Source: Le Figaro
