CHICAGO (AP) — Social media platform X is making changes to its AI chatbot after five secretaries of state warned it was spreading misinformation about the election.
The top election officials in Michigan, Minnesota, New Mexico, Pennsylvania and Washington state sent a letter to Elon Musk this month saying the platform's AI chatbot, “Grok,” generated misinformation about their states' voting deadlines shortly after President Joe Biden dropped out of the 2024 presidential race.
The secretaries of state have asked that the chatbot direct users with election-related questions to CanIVote.org, a voting information website run by the National Association of Secretaries of State.
The chatbot now states, “For accurate, up-to-date information about the 2024 U.S. elections, visit Vote.gov,” before listing answers to election-related questions.
The two websites are “trusted resources connecting voters with local election officials,” the five secretaries of state said in a joint statement.
“We appreciate Company X's efforts to improve its platform and look forward to them continuing to do so to ensure users have access to accurate information from trusted sources during this critical election year,” they said.
Grok is only available to X's premium subscribers. But the five secretaries of state who signed the letter said election misinformation from Grok was being shared across multiple social media platforms, reaching millions of people. Grok repeated the misinformation for 10 days before correcting it, the secretaries said. The platform did not respond to requests for comment.
The change to promote links to official voting websites does not appear to address Grok's ability to create misleading, AI-generated images related to the election. People have used the tool to flood the platform with fake images of candidates, including Vice President Kamala Harris and former President Donald Trump.
Grok debuted last year for X Premium and Premium Plus subscribers and was touted by Musk as a “rebellious” AI chatbot that would answer “tough questions that most other AI systems would refuse to ask.”
Social media platforms have faced increased scrutiny for their role in spreading misinformation, especially around elections. The letter also warned that inaccuracies are to be expected in AI products, especially chatbots such as Grok that are based on large-scale language models.
Since Musk bought Twitter in 2022 and renamed it X, watchdog groups have expressed concern about a surge in hate speech and misinformation on the platform and cuts to staff who moderated content.
The story continues
Experts say the move marks a step back from progress made by social media platforms in combating political disinformation after the 2016 US presidential election, and could lead to a worsening disinformation landscape ahead of this November's election.
___
The Associated Press receives support from several private foundations to strengthen its commentary coverage of elections and democracy. Learn more about the AP Democracy Initiative here. The Associated Press is solely responsible for all content.
