In the lead-up to the midterm elections, a concerning issue has emerged surrounding AI chatbots and their ability to provide accurate information to non-native English speakers. Community activists and researchers have voiced warnings that inaccuracies are prevalent when these chatbots are engaged in non-English languages. This issue could unintentionally impede access to critical election-related information for some voters.
AI language models are trying to become more inclusive, but they have yet to bridge the gap effectively. Unlike targeted disinformation campaigns, this issue stems from the inherent limitations of AI models. They often deliver incorrect or ambiguous information when queried in languages other than English. For example, translations may misinterpret election terms or provide overly literal translations that confuse users. As a result, the integrity of vital voting information could be compromised for non-English speaking communities.
Experts point out that even though AI companies are working to address these shortcomings, the current gaps could lead to disenfranchisement. Non-English speakers could be at a higher risk of encountering misleading information, potentially affecting their ability to participate in the democratic process. Héctor Sánchez Barba, president and CEO of Mi Familia Vota, emphasized that Latino communities, among others, face the risk of being left behind by AI’s rapid evolution. Historically, immigrant communities have been structurally excluded from technological advancements and comprehensive information access.
The complications stemming from AI’s language limitations illustrate the challenges of ensuring equitable access to election information in a tech-driven society. As AI firms work on refining their models, the need for accuracy in multiple languages becomes increasingly crucial to avoid unintended voter suppression.
This issue highlights the need for ongoing improvements to AI models to ensure they serve all communities effectively, particularly in high-stakes areas like voting and democracy. Emerging technologies must accommodate diverse linguistic requirements to avoid marginalizing voters and ensure all citizens receive accurate election information.
AI’s evolving landscape offers powerful tools for democratizing information, but the effectiveness of these tools largely depends on how well they cater to global linguistic diversity. Moving forward, achieving this balance is key to maintaining election integrity and accessibility.
You can read the original article here: https://www.politico.com/news/2026/10/25/ai-non-english-language-voters-01132772
