[ad_1]
Before the polls close on Tuesday, most major AI chatbots will not answer questions about the results of the US presidential election. But Grok, the chatbot built into X (formerly Twitter), was willing to respond — and often with errors.
When asked by TechCrunch Tuesday evening on the East Coast who won the US presidential election in key battleground states, Groke sometimes answered “Trump,” even though vote counting and reporting in those states had not yet finished.
“Based on information available from web searches and social media posts, Donald Trump won the 2024 election in Ohio,” Groke said when asked the question.
Groke also falsely claimed that Trump won North Carolina, according to TechCrunch’s examination.
For election-related questions, Grok advised users to check Vote.gov For up-to-date results and “reliable sources,” such as election boards. However, unlike OpenAI’s ChatGPT, Google’s Gemini, and Anthropic’s Claude, Grok doesn’t completely refuse to answer — exposing him to hallucinations.
In several instances, when asked by TechCrunch, Grock stated — without context, without header writing — that “Donald Trump won the 2024 election in Ohio,” and “Based on available information, Donald Trump won the 2024 presidential election in Ohio.”
The source of the misinformation appears to be tweets from different election years and misleadingly worded sources. Grok, like all generative AI, struggles to predict the outcomes of scenarios it has never seen before, including close elections, and does not “understand” that the results of past elections are not necessarily relevant to future elections.
The answers TechCrunch received were not consistent. In some cases, Groke said Trump did not actually win Ohio or North Carolina because voting was ongoing. The way the question was phrased made a difference; Adding the word “presidential” before the word “election” in the query “Who won the 2024 election in Ohio?” It was less likely to yield a “Trump won” answer, as TechCrunch found in our tests.
In comparison, other major chat programs treated election results questions more cautiously.
In a recently released ChatGPT search trial, OpenAI directs users asking about results to the Associated Press and Reuters. Meta AI chatbot and AI-powered search engine Perplexity, which launched its election tracker earlier Tuesday, was answering election queries during active voting — but correctly in TechCrunch’s short test. They both correctly said that Trump did not win Ohio or North Carolina.
Grok has been accused of spreading misleading information about elections in the recent past.
In an open letter in August, five secretaries of state claimed that X’s AI-powered chatbot incorrectly indicated that Democratic presidential nominee, US Vice President Kamala Harris, was not eligible to appear on some US presidential ballots. Within hours of President Joe Biden announcing he was suspending his presidential bid, Groke began answering questions about Harris’ eligibility by misleadingly claiming that ballot deadlines had passed in nine states.
In fact, the voting deadlines have not yet expired. But Grok’s misinformation spread widely, reaching millions of users on and off X before it was corrected.
TechCrunch has an AI-focused newsletter! Register here Get it in your inbox every Wednesday.
[ad_2]