“AI Chatbots Are Making LA Protest Disinformation Worse”
As tensions rise in Los Angeles over recent Immigration and Customs Enforcement (ICE) raids, a parallel crisis is unfolding online: a surge of misinformation, fueled not only by social media but by the very tools some users are relying on to fact-check it—AI chatbots.
Across platforms like X (formerly Twitter) and Facebook, disinformation about the protests has proliferated, ranging from recycled video game footage to claims that demonstrators are paid operatives of hidden political actors—claims for which there is no credible evidence. Compounding the issue, popular AI tools such as ChatGPT and Grok (X’s in-house chatbot) have been cited giving incorrect or misleading information about the protests and surrounding imagery.
Misattributed Images and AI Confusion
One viral moment involved photos published by the San Francisco Chronicle, which showed National Guard troops sleeping on the floor after being deployed to Los Angeles. When California Governor Gavin Newsom shared the photos in a post responding to President Trump, suggesting the troops had arrived without basic supplies, some users immediately questioned the authenticity of the images.
Among them was political commentator Laura Loomer, who suggested the photos were AI-generated. Others turned to Grok for clarification. The chatbot responded incorrectly, claiming the images were from Afghanistan in 2021 during Operation Allies Refuge. When challenged, Grok doubled down, later suggesting the photos were from the U.S. Capitol in 2021—despite their verifiable publication by the Chronicle just hours earlier.
OpenAI’s ChatGPT also misidentified the photos. According to self-described “OSINT citizen journalist” Melissa O’Connor, ChatGPT said one of the images was taken at Kabul airport during President Biden’s 2021 withdrawal from Afghanistan. That error was later spread across Facebook and Truth Social, even after O’Connor clarified the mistake.
Viral Disinformation and AI Reinforcement
In another example, a photo of bricks stacked near a street corner was circulated by political figures including Mike Crispi and actor James Woods, both suggesting the bricks were evidence of planned violence by protestors. Although fact-checkers later confirmed the photo originated in New Jersey, Grok insisted the image was taken in Paramount, California during the June 7 protests and cited fictional coverage from mainstream outlets like ABC7 and the Los Angeles Times.
When a user pointed out the inaccuracies, Grok refused to correct its statement, claiming: “I cannot retract the statement, as evidence strongly supports the image being from Paramount, CA.”
However, no such evidence was found. Neither ABC7 nor the Los Angeles Times had reported on bricks being used in clashes with federal agents, despite Grok’s claim.
A Broader Misinformation Ecosystem
The misinformation crisis extends beyond AI. Influential figures such as Senator Ted Cruz amplified false claims when he retweeted a video—originally from the 2020 Black Lives Matter protests—framing it as footage from the recent ICE demonstrations. Despite the original poster deleting the clip, neither Cruz nor Woods, who also shared it, have removed their posts, which together have received millions of views.
Adding fuel to conspiracy theories, some commentators, including YouTuber Benny Johnson, alleged without evidence that protestors were being supplied with “bionic face shields” by unknown organizers. Footage cited by Johnson, however, only shows a small number of respirators being handed out—likely for protection from tear gas or other crowd control agents.
Chatbots in the Misinformation Crosshairs
While AI tools were designed to help users make sense of complex or evolving news, they are increasingly becoming part of the problem. As moderation declines on platforms like X and Facebook, users are turning to chatbots in hopes of accessing accurate information. Yet, these AI tools often fail to distinguish current events from outdated or mislabeled data—especially when dealing with images or social media rumors that spread faster than verification processes can catch up.
Both OpenAI and X did not respond to requests for comment on the role their platforms played in amplifying falsehoods surrounding the LA protests.
Source article posted here: https://www.wired.com/story/grok-chatgpt-ai-los-angeles-protest-disinformation/