In This Story
Five of the U.S.’s top election officials on Monday asked X owner and chief technical officer Elon Musk to ensure his artificial intelligence chatbot stuck to the facts after it spread false information regarding Vice President Kamala Harris.
Minnesota Secretary of State Steve Simon sent Musk a letter Monday calling for X to “immediately” make changes to X’s Grok chatbot to prevent election-related misinformation from confusing voters. Simon was joined by his counterparts in New Mexico, Michigan, Washington, and Pennsylvania, who co-signed the letter and collectively oversee 37 million constituents.
“As tens of millions of voters in the U.S. seek basic information about voting in this major election year, X has the responsibility to ensure all voters using your platform have access to guidance that reflects true and accurate information about their constitutional right to vote,” the officials state in the letter.
The letter comes in response to false information provided by Grok in the aftermath of President Joe Biden’s decision to drop his re-election campaign on July 21 and endorsed Harris as the Democratic nominee. Within hours of his announcement, Grok generated a post that informed users that the ballot deadline for at least nine states had passed, meaning that he cannot be replaced as the Democratic party’s pick.
However, that information was completely inaccurate and ballots in each of the nine states mentioned were not closed; in fact, it had not passed in any states as of July 30, Reuters reported. All five states represented by the officials who signed the letter were mentioned by Grok as having closed ballots.
Although the secretaries of state note that only X Premium and Premium+ subscribers have access to Grok, that information can be shared by users that pay for access to the chatbot. “Millions of people” have seen the false information, which was repeated for more than a week without correction, according to the letter.
One remedy suggested by the secretaries of state is that, when users ask to see election-related information, they are directed to CanIVote.org, a website hosted by the National Association of Secretaries of State. OpenAI’s ChatGPT has already partnered with the association.
Grok was created by Musk’s AI startup xAI, and launched on X in April. The chatbot’s AI news summaries replace X’s trending section for paying users. Although Musk has called the bot “anti-woke” and promised it will be “truth-seeking” and unfiltered, it’s run into issues for spreading fake information.
Grok has been wrong about everything from a solar eclipse to New York City Mayor Eric Adams supposedly sending 50,000 police officers into the city’s subway. It also cited parody news website The Onion for a post following the death of football player O.J. Simpson.
On the more harmful side, Grok has claimed that the individual who attempted to assassinate former president Donald Trump was a member of Antifa, a loose network of far-left individuals, and that Harris was the one who was actually shot. Each of Grok’s headlines and summaries are accompanied by a small disclaimer, which says, “This story is a summary of posts on X and may evolve over time. Grok can make mistakes, verify its outputs.”