Elon Musk's AI App X Spreads Voter Fraud Theories via Unverified Claims
Image credits: Freemalaysiatoday

Elon Musk’s AI App X Spreads Voter Fraud Theories via Unverified Claims

Elon Musk’s AI-powered app X accelerates the spread of voter fraud theories by displaying unproven claims in its “explore” section.

With the help of AI, Elon Musk’s social media app X is speeding up the spread of voter fraud conspiracy theories. It is doing this by disseminating false claims, such as two personal attacks on Vice President Kamala Harris. The app’s “explore” section shares that it uses Musk’s AI software, Grok, to collect popular social media topics. Real people don’t seem to check the data, and in a few recent cases, the data appeared to repeat false or unproven claims as true.

Not all users can access “Stories for you” because it’s in beta testing. There is a feed of posts about a popular topic in each “story for you.” When users access the “story for you” history in the desktop version of X, the Grok software provides a paragraph-long summary of the subject matter.

The feature’s placement in X’s explore section makes it stand out in the digital world during the last few weeks of the presidential election. Musk is supporting former President Donald Trump. Musk has been repeatedly disseminating false information and conspiracy theories about the election. This comes after Musk has shared similar ideas in public and on his social media.

NBC News found five “stories for you” in the last week that made false claims about the election.  Grok selects popular topics, but each one carries a warning stating that Grok isn’t accountable for its accuracy, and users should independently verify the facts: The disclaimer says, “Grok can make mistakes; check its outputs.”

On Monday, Grok repeated without question false claims of wrongdoing involving the voting machine company Dominion Voting Systems. Grok put together a “story for you” called “Public Scrutiny of Dominion Voting Systems.” Posts in the story accused the company of “rigging elections” and “fraud.”

Dominion has said that similar claims were false in the past, and Fox News agreed to pay Dominion $787.5 million to settle a defamation lawsuit it filed last year.  In his written summary of the online debate, Grok said that Dominion was using “legal threats” to “potentially silence legitimate discussions on election security.”

On Wednesday, Grok repeated false accusations of wrongdoing in Maricopa County, Arizona. He supported an X user’s claim that election workers in the county are “corrupt” because of how quickly they count ballots.

This week, Grok spread the false claim that a voting machine in Tarrant County, Texas, was “flipping” votes. This is the third election-related “story for you.” X’s AI software pushed posts claiming multiple examples, but only one voter came forward, and his claim remained unchecked.

Officials in the area said there was no proof of what he said, which makes it possible that he pressed the wrong candidate by accident. They say that after he looked over his ballot, he was able to cast the vote he meant to.

Grok has also said negative things about Harris, who is running for president as a Democrat. They created a “story for you” that echoed false allegations from X users, claiming that Harris used cocaine in the White House and attended parties hosted by Sean “Diddy” Combs, who is currently facing federal sex-abuse charges. Fact-checkers claim that they added Harris’ face to a photo of Combs with fashion designer Misa Hylton.

Numerous individuals have viewed some of the posts that Grok curates. Whether most of those views occurred before or after Grok made the posts trending is unclear. Musk, the CEO of Tesla and SpaceX, has propagated some of the same conspiracy theories that his Grok software has reinforced. For example, he has made false claims that Dominion Voting Systems did something wrong.

According to the Bloomberg Billionaires Index, Musk is the richest person in the world. He has donated almost $75 million to a superPAC that supports Trump, and he has traveled all over Pennsylvania, a battleground state, to get pro-Trump voters registered and to the polls, using a $1 million giveaway every day to get people’s attention.

NBC News asked X to say something about its “stories for you” feature. X received a list of emails containing questions, but no one responded. X received a list of emails containing questions, but no one responded.

A representative for Dominion Voting Systems did not have an immediate comment on X’s “stories for you” feature. However, they did point to a website that defends the accuracy of the company’s systems. The Harris campaign did not respond when asked for comments.

People have encountered issues with this feature in the past. In July, Grok’s summaries helped spread false information about President Joe Biden. These summaries fostered fanciful conspiracy theories, suggesting that Biden was concealing a medical emergency, might be in danger of death, or had already met his demise.

Up until just two years ago, when Musk bought the app that was then known as Twitter, curators performed this function. The people in charge of Twitter added curated trending topics in 2015 to help users make sense of all the information on the app.

To improve the quality of popular information, Twitter employees added verified context from well-known news sources like The Associated Press and Reuters.

However, one of Musk’s first actions after purchasing Twitter was to eliminate the positions of human curators. Now, the tech billionaire has given the job to his AI software.

X’s AI-Written Summaries Face Criticism 

Musk often criticizes traditional news outlets like The Associated Press and Reuters. Despite its numerous factual errors, debunked claims, and conspiracy theories, he has also promoted X as the internet’s most trustworthy news source.

X initially tested AI-written summaries months ago. The experiment hasn’t gotten much attention, but some users were quickly worried about the Grok-written text they were seeing in the beta test of the app’s “explore” section.

In April, a Reddit user said that the summaries of popular topics were like rolling “brainless AI dice.” Someone on X also posted a screenshot of the “explore beta” in July, containing text indicating that Biden would take over as the next coach of England’s national soccer team.

Many X users have written about the “explore beta” in the last few weeks, and the majority of them have been negative. Someone said it had no point, and someone else said it was “absolute trash.” Some said they couldn’t leave the beta test easily.

It’s not clear how many people see the summaries made by AI. Some X users, who browse through the app’s trending topics, are only presented with ranked lists of popular topics. They don’t encounter any AI-written text. It is also unclear if X plans to make the Grok summaries a permanent part of the app or just a beta test.

 

Leave a Reply

Your email address will not be published. Required fields are marked *

Meta AI Launches Token-Level Detective Reward Model for Improved Vision Language Model Accuracy Previous post Meta AI Launches Token-Level Detective Reward Model for Improved Vision Language Model Accuracy