As nearly half of US adults have used AI chatbots and some rely on them for election insights, recent analyses reveal that these tools frequently deliver incorrect, outdated, or partisan-tainted information about candidates and voting details.
- Almost 40% of election-related chatbot answers include factual errors.
- Many chatbots cite politically biased ‘pink slime’ news sites nearly half the time.
- Voters risk misinformation if relying on AI chatbots for candidate or voting info.
What happened
Recent research by Forum AI and NewsGuard reveals that mainstream AI chatbots, including OpenAI's ChatGPT, Anthropic's Claude, and Google's Gemini, have significant shortcomings in delivering election-related information. Approximately 40% of answers about candidates and voting guidelines from these tools contained factual inaccuracies. Instances include Claude confusing a sitting congressman with a former school superintendent and Gemini giving incorrect ballot request deadlines.
Moreover, these chatbots frequently pull information from politically motivated sites disguised as mainstream news sources, known as 'pink slime' websites, which distort or spin facts to align with ideological agendas. ChatGPT was notably the worst offender, citing such biased sites in over 70% of inquiries related to the 2026 midterms.
Why it matters
AI chatbots have become a common resource for voters seeking quick political information, according to Pew Research. With 50% of US adults ever using AI chatbots and 8% leveraging them specifically for midterm election insights, the risk of misinformation is amplified. Erroneous data can affect voter knowledge, choices, and confidence during critical democratic moments.
The problem stems from the underlying data sources feeding chatbots and their reliance on pattern prediction rather than verified facts. Given the explosion of partisan websites pushing misleading narratives, AI responses can propagate these biases unintentionally, misleading users who may not have the skills or awareness to detect such distortions.
What to watch next
As the 2026 US midterms progress, attention will focus on how AI developers improve chatbot accuracy and mitigate partisan biases. OpenAI has outlined efforts to address election misinformation, though transparency and third-party verification remain limited. Monitoring how AI platforms adapt and communicate their safeguards will be key to determining their future utility in political information dissemination.
Voters and policymakers should also watch for initiatives that raise public awareness about AI's limitations in election contexts. Increasing literacy around AI-generated content and encouraging cross-verification with trusted, independent sources will be crucial to minimizing reliance on flawed chatbot outputs in the democratic process.