Dutch Watchdog Warns Against AI Chatbots for Voting Advice

2025-10-22 data

Amsterdam, Wednesday, 22 October 2025.
The Dutch Data Protection Authority cautions against AI chatbots for voting advice, highlighting their bias and unreliability, which could distort democratic processes ahead of the 29 October election.

Concerns Over AI Influence on Democratic Integrity

The Dutch Data Protection Authority (Autoriteit Persoonsgegevens, AP) has raised significant concerns about the use of AI chatbots, such as ChatGPT, Gemini, Mistral, and Grok, for voting advice. These tools are criticized for providing unreliable and biased recommendations, often favoring the far-right Freedom Party (PVV) and the left-wing Green Left-Labour Party (GroenLinks-PvdA) in more than half of the cases examined [1][2]. Monique Verdier, the AP’s deputy chair, emphasized that such biases undermine the integrity of free and fair elections, a cornerstone of democratic societies [1].

Potential Biases in AI Chatbot Recommendations

Research conducted by the AP highlights that AI chatbots frequently direct users towards political fringes, with inconsistent recommendations even when user inputs align more closely with moderate parties. This phenomenon is attributed to the non-transparent and unverifiable nature of the data these chatbots are trained on [3][4][5]. The report underscores the risk of voters being nudged towards parties that do not reflect their actual preferences, potentially distorting electoral outcomes [2][5].

The Call for Human Judgment in Voting

As the Netherlands approaches its national elections on 29 October 2025, the AP urges voters to rely on traditional methods of gathering political information, such as reading party platforms and watching debates, rather than depending on AI-generated advice [3][6]. The authority’s cautionary stance reflects a broader concern about the increasing role of technology in critical decision-making processes, where human judgment and discernment are deemed irreplaceable [1][6].

Future Implications for AI in Electoral Contexts

The AP’s warnings also call for developers to implement measures that prevent AI chatbots from being used as voting aids, emphasizing the need for transparency and accountability in AI systems. This situation highlights a growing dialogue on how technology can be integrated responsibly into democratic processes without compromising their integrity [4][7]. As the field of AI continues to evolve, ensuring that these tools are used ethically and effectively remains a critical challenge for regulators and technology developers alike [7].

Bronnen

artificial intelligence voting advice