Election voting advice from AI chatbots ‘inaccurate and unreliable’
Research during Hungary election shows AI recommended parties not running and gave highly volatile answers to identical prompts AI chatbots provide inaccurate, inconsistent and unreliable guidance to voters asking which party they should back, a study...
AI chatbots can give voters inconsistent, inaccurate and sometimes plainly misleading guidance on election choices, according to a study based on research carried out during Hungary’s parliamentary elections this year.
The findings, highlighted in a report monitored from a public feed, suggest that general-purpose AI systems may be a poor source of electoral advice. In tests, the chatbots reportedly recommended parties that were not even taking part in the election, failed to mention the correct party in some responses, and produced highly volatile answers when researchers repeated the same prompt.
The study, published by the civil liberties group Liberties, says those results raise “serious concerns” about the reliability of AI tools in electoral settings. Rather than offering stable or helpful information, the systems appeared to vary sharply from one response to another, underscoring the risks of treating chatbot output as neutral or dependable political guidance.
The research adds to growing scrutiny of how AI tools behave in public-interest contexts, especially when users turn to them for fast explanations on complex topics such as voting. Unlike a curated voter guide or an official election resource, a general-purpose chatbot may generate answers based on patterns in its training data and live retrieval systems, which can produce errors or inconsistencies.
According to the feed summary, the issue was not limited to occasional mistakes. Researchers found that the same prompts could lead to different answers, and those answers could include parties that had no role in the contest. That kind of instability could be particularly problematic for voters seeking straightforward advice on which party aligns with their views.
The Hungarian case offers an example of how AI systems can fall short when asked to handle civic information that demands precision. Even a small error in a voter-facing answer can distort political understanding, and repeated inconsistency may make it difficult for users to know whether any given response can be trusted.
The report does not appear to argue that AI has no place in election information at all, but it clearly warns against relying on broad-purpose chatbots as electoral advisers. Its central message is that voters need accurate, stable and transparent sources, and that current AI systems may not reliably provide them.
Source: The Guardian Technology - https://www.theguardian.com/technology/2026/jul/21/election-voting-advice-ai-chatbots-inaccurate-unreliable-hungary
