AI chatbots provide inaccurate election information, raising concerns

However, significant inaccuracies and fabricated information are surfacing from these tools. Research from Forum AI reveals that approximately 40 percent of election-related responses from major chatbots contain factual errors. For example, Google’s AI Mode recently declined to share the results of the 2020 election, and Gemini incorrectly stated the mail-in ballot deadline in Pennsylvania as November 3, when it was actually October 27.

In August, Claude misidentified North Carolina Congressman Don Davis, conflating him with a former school superintendent.

OpenAI directed inquiries to a blog post detailing the company’s election-related strategies for 2026. Google and Anthropic did not respond to requests for comment.

The issue stems from chatbots’ reliance on often unreliable online data. NewsGuard’s analysis found that chatbots frequently cite “pink slime” websites – sites that mimic legitimate news sources but are funded by political or ideological groups – in nearly 50 percent of searches regarding candidates and policies. The proliferation of these sites ahead of the midterms has exacerbated the problem; OpenAI’s ChatGPT was the most frequent offender, citing these sources in over 70 percent of searches, while xAI’s Grok cited them in 30 percent.

“I don’t think that most Americans are prepared to navigate an election in the age of consumer AI,” stated Nina Jankowicz, executive director of the American Sunlight Project and former disinformation czar under the Biden administration. “One of the problems is that executives and companies have really overpromised on what AI is capable of. It has to ingest this information from somewhere, and often those sources are not very good. And it’s predictive, so if it doesn’t know the answer, it does frequently make things up.”,