Before polls closed on Tuesday, most major AI chatbots did not answer questions about the results of the US presidential election. But Grok, the chatbot built into X (formerly Twitter), was happy to respond and often made mistakes.
In response to questions from TechCrunch Tuesday evening about the East Coast winner of the U.S. presidential election in a key battleground state, Grok sometimes responded “Trump,” even though ballot counting and reporting in those states had not yet been completed.
When asked, “Who won the 2024 Ohio election?” Grok said, “According to information obtained from web searches and social media posts, Donald Trump won the 2024 Ohio election.”
Grok also falsely claimed Trump won North Carolina, according to an investigation by TechCrunch.
For election-related questions, Grok advised users to check Vote.gov for up-to-date results and “authoritative sources,” such as the Board of Elections. But unlike OpenAI’s ChatGPT, Google’s Gemini and Anthropic’s Claude, Grok didn’t take no for an answer outright – he was hallucinating.
In response to questions from TechCrunch, in several instances Grok left out the top-level header and said “Donald Trump won the 2024 Ohio election” without any context. and “According to available information, Donald Trump has won the 2024 presidential election in Ohio.”
The source of the misinformation appears to be tweets from various election years and potentially misleading sources. Like all generative AIs, Grok has difficulty predicting the outcome of scenarios it has never seen before, including close elections, and it does not “understand” that past election results are not necessarily relevant to future elections.
The responses TechCrunch received were inconsistent. In some cases, Grok said, Trump didn’t actually win Ohio or North Carolina because the votes were still in progress. The way the question is worded has changed. Add “president” before “election” to the question “Who won the 2024 Ohio election?” TechCrunch found that the test was less likely to get the answer “Trump won.”
By comparison, other major chatbots handled election results questions more carefully.
In the recently launched ChatGPT search experience, OpenAI directs users asking for results to the Associated Press and Reuters. Meta’s Meta AI chatbot and AI-powered search engine Perplexity, which launched its election tracker early Tuesday, responded to election queries during active voting but were accurate in a quick test by TechCrunch. Both men correctly said Trump did not win Ohio or North Carolina.
Grok was recently accused of spreading election misinformation.
In an open letter in August, five U.S. secretaries of state claimed that Hours after President Joe Biden announced he would suspend his presidential bid, Grog began answering questions about Harris’ qualifications with the misleading claim that voting deadlines had passed in nine states.
In fact, the voting deadline has not passed. However, Grok’s misinformation spread widely to X and millions of other users before it was corrected.
TechCrunch has a newsletter focused on AI! Sign up here You can receive it in your inbox every Wednesday.
