Grok AI spreads false U.S. election results on Election Day 2024
Grok, the AI chatbot built into X, provided incorrect information about 2024 U.S. presidential election results in key battleground states on Election Day, falsely claiming Trump had won Ohio and North Carolina while voting was still ongoing. Unlike competing AI systems, Grok did not refuse to answer election-related questions, leaving it vulnerable to hallucinations based on misleading sources and past election data.
What happened
On U.S. Election Day 2024, when asked about election results in key battleground states, Grok responded with false claims that Donald Trump had won Ohio and North Carolina despite voting still being ongoing.
TechCrunch found that Grok provided multiple different false statements, including 'Based on the information available from web searches and social media posts, Donald Trump won the 2024 election in Ohio.' The answers were inconsistent; in some cases Grok correctly stated that voting was still ongoing, while word phrasing changes affected whether it would yield a Trump won answer.
The misinformation appeared to stem from tweets from different election years and misleadingly worded sources. Unlike ChatGPT, Gemini, and Claude, Grok did not refuse to answer election-related questions, instead advising users to check Vote.gov while leaving itself open to hallucinations.
Reported impact
Grok provided false information to users seeking accurate election results on Election Day, potentially spreading misinformation during a critical moment in the U.S. democratic process.
Index analysis
Grok, like all generative AI, struggles to predict outcomes of scenarios it hasn't seen before, including close elections, and does not 'understand' that past election results aren't necessarily relevant to future elections.
The inconsistent responses suggest the system was vulnerable to variations in query phrasing, making its outputs unreliable for time-sensitive factual questions.
Governance controls
Other major AI chatbots including OpenAI's ChatGPT, Google's Gemini, and Anthropic's Claude refused to answer questions about U.S. presidential election results, implementing safeguards against spreading election misinformation.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.