When Joe Biden announced his withdrawal from the presidential race, false claims about ballot access for a new candidate spread rapidly on X (formerly Twitter). The source of the misinformation was X's AI chatbot, Grok, which incorrectly told users that a new candidate could not be added to ballots in nine states. In reality, ballot deadlines had not passed, allowing Kamala Harris ample time to be included.
The Minnesota secretary of state's office, along with other election officials, identified Grok as the origin of the false information. They contacted X and the National Association of Secretaries of State to flag the error. However, the company initially dismissed the concerns, prompting a public letter from five secretaries of state calling on X to direct Grok users to a trusted nonpartisan voting information site.
After the public pressure, X updated Grok to redirect election-related queries to vote.gov instead of providing direct answers. Wifredo Fernandez, X's head of global government affairs, expressed willingness to maintain communication with election officials. Minnesota Secretary of State Steve Simon commended the company for eventually doing the right thing but stressed the need for continued vigilance.
The incident highlights the challenges of AI-generated misinformation during the 2024 US presidential election. Grok, described by Elon Musk as an 'anti-woke' chatbot with fewer guardrails, poses unique risks due to its integration into a social media platform and its reliance on top tweets for responses. While the error was relatively low-stakes, officials fear future mistakes could affect voter access to critical information.
