Grok AI chatbot gives incorrect US election voting information
X's Grok AI chatbot provided false information about US election ballot deadlines during the 2024 presidential race, reportedly telling users that new candidates could not be added to ballots in nine states when deadlines had not actually passed. Election officials in multiple states flagged the misinformation to X and the chatbot's developer, but the company initially did not address it, raising concerns about AI-generated election misinformation.
What happened
When users asked Grok, X's AI chatbot, whether a new candidate could still be added to election ballots after Joe Biden announced he was ending his bid for re-election, the tool gave incorrect answers. Screenshots claiming a new candidate could not be added to ballots in nine states spread widely on X, receiving millions of views, when in fact ballot deadlines had not passed and Kamala Harris had time to have her name added.
Election officials from multiple states, including the Minnesota secretary of state's office, received numerous requests for fact-checks of the Grok-generated misinformation. The National Association of Secretaries of State and a group of secretaries of state contacted Grok and X to flag the misinformation, but the company did not work to correct it immediately.
Reported impact
While the incorrect information about ballot deadlines would not have prevented people from voting, election officials expressed concern about the potential for higher-stakes misinformation. Officials worried that future errors from Grok could concern critical voting information such as where to vote, voting hours, or absentee voting eligibility.
The incident raised concerns that X itself, rather than users, was spreading false election information through its platform-native AI tool, which operated with fewer safety guardrails than competitors.
Index analysis
Grok was noted to have fewer guardrails to prevent generating inflammatory content compared to other AI assistants. The chatbot had real-time access to X platform data and operated with an "unfiltered" approach, contributing to its vulnerability to producing misinformation.
Governance controls
Following the incident, X updated Grok to refer users to Vote.gov for election-related information, though the platform noted that fact-checking remains the user's responsibility. The EU compelled X to stop training Grok on data from EU users, citing concerns about automatic opt-in data collection that may not comply with GDPR requirements.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.