Before the polls closed on Tuesday, many popular AI chatbots, like those from OpenAI and Google, were cautious and refused to answer questions regarding the results of the U.S. presidential election. However, Grok, the AI chatbot integrated into X (which used to be known as Twitter), was willing to respond to such questions, but frequently provided incorrect information.
For example, when TechCrunch asked Grok who had won key battleground states in the presidential election, it sometimes mistakenly answered “Trump,” even though the vote counting in those areas had not finished yet. When asked specifically about Ohio, Grok declared, “Based on the information available from web searches and social media posts, Donald Trump won the 2024 election in Ohio,” despite the fact that the results were still in the process of being determined.
Grok also incorrectly stated that Trump won North Carolina, as confirmed by TechCrunch’s follow-up checks.
For questions related to the election, Grok recommended that users visit Vote.gov for the latest updates and refer to reliable sources like election boards. Unlike other chatbots such as ChatGPT, Google’s Gemini, and Anthropic’s Claude, which outright declined to provide answers to similar questions, Grok chose to answer—leading to errors in the information given.
In several interactions with TechCrunch, Grok claimed without any clarifying context that “Donald Trump won the 2024 election in Ohio” and repeated similar assertions based on the details it found online. The misinformation seemed to arise from tweets made during previous election years and from sources that were misleadingly worded. Grok, like many AI chatbots, struggles with real-time or close election scenarios because it can’t correctly draw distinctions between past and present election information.
Read Also: Nvidia Is Now The World’s Biggest Company Thanks To The AI Boom
The responses from Grok were inconsistent; at times, it correctly noted that Trump had not yet won Ohio or North Carolina because voting was still ongoing. Interestingly, the phrasing of the questions influenced the answers—adding the word “presidential” to the question “Who won the 2024 election in Ohio?” made it less likely to result in the answer “Trump won.”
In contrast, other prominent AI chatbots were handling inquiries about election results more carefully. For instance, OpenAI’s newest chatbot feature, ChatGPT Search, directed users looking for election results to credible news sources like The Associated Press and Reuters. Similarly, Meta’s Meta AI chatbot and Perplexity’s AI-powered search engine, which had launched an election tracker earlier that day, provided accurate answers, stating that Trump had not won in Ohio or North Carolina during TechCrunch’s brief testing.
Grok has previously faced criticism for spreading false information about elections. In August, five state secretaries of state wrote an open letter stating that Grok had incorrectly suggested that Kamala Harris, the Democratic presidential candidate and current Vice President, was ineligible for some presidential ballots. Soon after President Joe Biden announced he would suspend his presidential campaign, Grok misleadingly claimed that deadlines had passed for ballots in nine states, sparking confusion that reached millions of users on X, though this misinformation was later corrected.
