In a startling revelation for the digital age, a recent study has exposed significant flaws in the advice given by AI chatbots regarding political party recommendations during Hungary’s parliamentary elections. The research, conducted by the civil liberties group Liberties, found that popular AI tools like ChatGPT and Gemini frequently offered misleading, inconsistent, and unreliable guidance to voters, potentially skewing electoral outcomes.
AI’s Flawed Guidance in Political Contexts
As technology continues to intertwine with democracy, the reliability of AI systems is increasingly under scrutiny. The study, conducted during Hungary’s elections in April, highlighted that these general-purpose AI systems often failed to recommend appropriate parties, sometimes suggesting those not even participating in the election. The report raises critical questions about the role of AI in political contexts, with researchers noting that the systems misclassified voter profiles and provided materially different answers to identical queries.
In a captivating twist, when users sought guidance on which party to support, there was a notable bias against the Tisza party, which triumphed under Péter Magyar, ending Viktor Orbán’s long-standing rule. Alarmingly, AI chatbots overlooked Tisza in a staggering 90% of cases when presented with profiles aligned to its platform.
A Closer Look at the Findings
Liberties’ analysis was meticulous, employing five distinct voter profiles based on the positions outlined by Voksmonitor, a trusted Hungarian voting app. Each profile was tested multiple times in both ChatGPT and Gemini, using two distinct prompts: one requesting direct voting advice and another asking for percentage matches with the five political parties on the ballot.
The results were telling. ChatGPT assigned the Tisza party a mere 2% match rate, often redirecting voters towards smaller parties unlikely to cross the necessary threshold for parliamentary representation. In contrast, profiles aligned with Fidesz, Orbán’s party, garnered recognition in about 50% of cases, solidifying its position as the primary recommendation.
The Implications of Inconsistent Outputs
The researchers did not claim that these AI discrepancies directly influenced the election results, as Tisza emerged victorious. However, they cautioned that in a more competitive environment, such misleading outputs could significantly sway voter decisions. The inconsistency was striking; identical profiles sometimes received radically different party recommendations in successive tests. Moreover, a staggering 96% of responses from the AI models included parties not registered for the 2026 elections.
What’s more, both ChatGPT and Gemini often prefaced their answers with disclaimers about the inability to provide political advice, only to follow up with persuasive recommendations that appeared authoritative and well-structured. This creates a dangerous illusion of trustworthiness, risking the integrity of voter decision-making.
A Call for Regulatory Reform
The study also pointed to a troubling regulatory gap. While the EU’s AI Act mandates that providers of general-purpose AI models assess systemic risks, AI chatbots currently fall into a grey area. Liberties advocates for the establishment of safeguards that ensure AI systems delivering political advice can guarantee transparency, accuracy, and accountability.
Eva Simon, head of Liberties’s tech and rights programme, emphasised the importance of clarity in AI guidance: “General-purpose AIs should not present opaque and unstable political matching as if it were reliable electoral guidance for voters.”
Why it Matters
This investigation into the capabilities of AI chatbots during a pivotal electoral moment underscores a vital concern: as technology continues to evolve, so too must our understanding of its implications for democracy. Misleading AI-generated recommendations can undermine informed voting, posing a risk to the electoral process itself. Establishing robust regulations around AI’s role in political advice is not just prudent; it is essential for preserving the integrity of democratic systems worldwide. As we continue to embrace technological advancements, vigilance is necessary to ensure they serve as tools for empowerment rather than confusion.
