Secretaries of state urge Musk to fix AI chatbot spreading false election info

Five secretaries of state plan to send an open letter to billionaire Elon Musk on Monday, urging him to “immediately implement changes” to X’s AI chatbot Grok, after it shared with millions of users false information suggesting that Kamala Harris was not eligible to appear on the 2024 presidential ballot.

The letter, spearheaded by Minnesota Secretary of State Steve Simon and signed by his counterparts Al Schmidt of Pennsylvania, Steve Hobbs of Washington, Jocelyn Benson of Michigan and Maggie Toulouse Oliver of New Mexico, urges Musk to “immediately implement changes to X’s AI search assistant, Grok, to ensure voters have accurate information in this critical election year.”

Within hours of President Biden’s announcement that he was suspending his presidential campaign on July 21, “false information on ballot deadlines produced by Grok was shared on multiple social media platforms,” the secretaries wrote.

The secretaries cited a post from Grok that circulated after Biden stepped out of the race: “The ballot deadline has passed for several states for the 2024 election,” the post read, naming nine states: Alabama, Indiana, Michigan, Minnesota, New Mexico, Ohio, Pennsylvania, Texas and Washington.

Had the deadlines passed in those states, the vice president would not have been able to replace Biden on the ballot. But the information was false. In all nine states, the ballot deadlines have not passed and upcoming ballot deadlines allow for changes to candidates.

“This latest episode is unfortunate, but it’s also an opportunity to deliver a collective warning about the need for action on behalf of America’s voters,” Simon said in a message to The Washington Post. “We are all united by the goal of ensuring that voters get accurate information — and that they seek out trusted sources for such information.”

A message to X seeking comment from Musk, who controls X, was not immediately answered.

Musk launched Grok last year as an anti-“woke” chatbot, professing to be frustrated by what he says is the liberal bias of ChatGPT. In contrast to AI tools built by OpenAI, Microsoft and Google, which are trained to carefully navigate controversial topics, Musk said he wanted Grok to be unfiltered and “answer spicy questions that are rejected by most other AI systems.”

The secretaries of state, who are the chief elections officers in their states, are objecting not to Grok’s tone but its factual inaccuracies and the sluggishness of the company’s move to correct bad information.

Secretaries of state are grappling with an onslaught of AI-driven election misinformation, including deepfakes, ahead of the 2024 election. Simon testified on the subject before theSenate Rules and Administration Committee last year.

Many of them are also still beating back the ramifications of widespread false conspiracy theories that plagued the last presidential election. The messages from Grok spurred a public conversation about whether Harris would be a legitimate candidate for president, even though she declared her candidacy well within the necessary state deadlines. Such false assertions are the kind of misinformation that helped fuel widespread beliefs in 2020 that the election was stolen from former president Donald Trump.

Another version of Grok’s false information about ballot deadlines included one telling users that ballots for the coming presidential election were already “locked and loaded.”

“So, if you’re planning to run for president in any of these states, you might want to check if you’ve already missed the boat,” the chatbot responded. “But hey, there’s always 2028, right?”

“It’s important that social media companies, especially those with global reach, correct mistakes of their own making — as in the case of the Grok AI chatbot simply getting the rules wrong,” Simon added. “Speaking out now will hopefully reduce the risk that any social media company will decline or delay correction of its own mistakes between now and the November election.”

Grok is available only to X Premium and Premium+ subscribers, but the false information about ballot deadlines was “shared repeatedly in multiple posts — reaching millions of people,” the letter read. Grok repeated false information for more than a week until it was corrected on July 31.

Simon expressed disappointment in the way X initially responded to the error. He said that the company’s response was “the equivalent of a shoulder shrug. Dismissive and detached.”

Simon reached out to his counterparts in all nine states mentioned in the Grok messages, but only four others agreed to sign the letter. All the signatories are Democrats, except for Schmidt of Pennsylvania,a Republican who is an appointee of Gov. Josh Shapiro, a Democrat and atop candidate to be Harris’s running mate.

The secretaries noted that this year, OpenAI partnered with the National Association of Secretaries of State to give voters correct election information, and ChatGPT has been programmed to direct users to CanIVote.org — a nonpartisan resource from professional election administrators of both major parties. Grok has entered into no such partnerships.

“We urge X to immediately adopt a policy of directing Grok users to CanIVote.org when asked about elections in the U.S.,” the letter concluded.

“This issue underlines the importance of checking with trusted sources of election information, such as your state or local election officials to get accurate information about the election process,” Toulouse Oliver said in a text message. “We do hope that X is able to address the issue with Grok reiterating this false information and hope that this can be prevented in the future.”

Benson confirmed her involvement in the letter-writing campaign. Spokespeople for the secretaries of state in Ohio and Indiana confirmed that they did not add their signatures to the letter. Representatives for those same offices in Alabama and Texas did not immediately respond to a request for comment.

Benson also confirmed that she had launched an investigation into a Musk-backed PAC, called the America PAC, which is supporting Trump. The investigation was first reported by CNBC.

Benson’s concern is the detailed voter information that the PAC collected from people living in Michigan and other battleground states through a section of the website that says “register to vote.”

When users in battleground states such as Michigan click on the “register to vote” tab on America PAC’s website, they can submit personal information but are not given a form to complete voter registration, CNBC reported. That contrasts with the experience of users in states that are not considered politically competitive, such as Wyoming or California. Those users can enter their email address and Zip code but are then directed to a voter registration page for their state, or back to the original sign-up section, according to CNBC.

Simon said that Musk’s AI chatbot cannot make the argument that it is simply facilitating different voices in the modern-day public square.

“This is a case where the owner of the public square (the social media company itself) is the one who introduced and spread the bad information — and then delayed correcting its own mistake after it knew that the information was false,” he said.