AI Chatbots Amplify Misinformation Amid Los Angeles Immigration Protests

AI Chatbots Are Making LA Protest Disinformation Worse

Disinformation about the Los Angeles protests is spreading on social media networks and is being made worse by users turning to AI chatbots like Grok and ChatGPT to perform fact-checking. As residents of the LA area took to the streets in recent days to protest increasingly frequent Immigration and Customs Enforcement (ICE) raids, conservative posters on social media platforms like X and Facebook flooded their feeds with inaccurate information. In addition to well-worn tactics like repurposing old protest footage or clips from video games and movies, posters have claimed that the protesters are little more than paid agitators being directed by shadowy forces -- something for which there is no evidence. In the midst of fast-moving and divisive news stories like the LA protests, and as companies like X and Meta have stepped back from moderating the content on their platforms, users have been turning to AI chatbots for answers -- which in many cases have been completely inaccurate. On Monday, the San Francisco Chronicle published images of National Guard troops sleeping on floors. They were later shared on X by California governor Gavin Newsom, who responded to a post from President Donald Trump by writing: "You sent your troops here without fuel, food, water or a place to sleep." Within minutes of the posts being shared, many users on X and Facebook were claiming that the images were either AI-generated or taken from a completely different situation. "Looks like @GavinNewsom used an AI photo to smear President Trump," conspiracist Laura Loomer alleged on X. Some users seeking clarity turned to X's own chatbot Grok, with one user to clarify where the photo was taken. "The photos likely originated from Afghanistan in 2021, during the National Guard's evacuation efforts in Operation Allies Refuge," Grok wrote. "Claims linking them to the 2025 Los Angeles deployment lack credible support and appear to be a misattribution. No definitive source confirms the Los Angeles connection." When challenged about the assertion by another X user who pointed out that the images were first obtained by the San Francisco Chronicle, Grok replied: "I checked the San Francisco Chronicle's claims. The photos of National Guard troops sleeping on floors are likely from 2021, probably the U.S. Capitol, not Los Angeles 2025." The San Francisco Chronicle did not immediately respond to a request for comment. Grok isn't the only tool potentially making the situation worse. Melissa O'Connor, who describes herself as an "OSINT Citizen Journalist", posted the results given to her by ChatGPT when she uploaded the pictures posted by Newsom of troops sleeping on the ground. OpenAI's chatbot incorrectly claimed that one of the pictures shared by the California governor was taken in Kabul airport in 2021 during former President Joe Biden's Afghanistan withdrawal. The same results were shared on other platforms such as Facebook and Trump's own Truth Social as evidence that the pictures were fake. (O'Connor subsequently posted to make clear she's now aware the photos are not four years old, though the original post was left up.)

5 misinformation trends that circulate around mass protests

Mass demonstrations are fertile ground for online misinformation. Credit: Mario Tama / Getty Images News via Getty Images Following days of protest decrying the actions of Immigration and Customs Enforcement (ICE) and militarized federal forces in Los Angeles, downtown L.A. residents remained under curfew this week while California governor Gavin Newsom went toe-to-toe with President Trump, who sent Marines into the area. Memes, and viral misinformation, were abundant. Meanwhile, the "No Kings" organizing coalition has planned thousands more peaceful demonstrations this weekend, specifically orchestrated to overlap with Trump's planned military parade. While the organizations behind the events are gearing up for what they've said will be the "largest single-day, peaceful protest" in the country's history, some worry they may be met with an equal show of state-sanctioned violence -- and attempts to undermine news coverage via viral falsehoods. Peter Adams, senior vice president of research and design for the News Literacy Project, explained that mass demonstrations are rife for online misinformation, patterns which can be spotted before protesters even touch ground. Adams runs the organization's online misinformation newsletter RumorGuard, which collates viral fact-checks and critical literacy tips. "Protests generally only happen around events that are highly polarizing and controversial," said Adams. "The more polarizing something is, the more people's political identities are activated and invoked. We know that these identities play a very strong role in how people react to information online. As polarization has increased in American society, this is even more at play." News and misinformation debunking organizations, such as bipartisan NewsGuard, have reputed dozens of false posts, including those made by conservative commentators alleging Newsom had posted an outdated or AI-generated photo of national guard troops as he criticized Trump's militarized protest response. Such posts were propelled by X chatbot Grok, which told some users that the photos, first published by the San Francisco Chronicle, were outdated -- ChatGPT users got similarly confusing responses. Others, including representatives of the Trump administration, fanned the flames with misattributed or outrightly false quotes from federal and international leaders, including California congresswoman Maxine Waters and Mexican president Claudia Scheinbaum. Viral posts from partisan voices on both sides of the conflict have attempted to sway, or overtly manipulate, public perception and recast reality in terms that are more favorable to their side, Adams explains. "Viral misinformation expresses patterns over time, and we can learn how to recognize those patterns and inoculate ourselves to a whole swath of viral rumors," explained Adams. As protests continue, here's what to look out for: "Chatbots often generate outdated or inaccurate details, and that can be especially true during fast-moving events," explained Adams and other News Literacy Project experts. While users are turning to built-in chatbots to fact-check on platforms that have removed internal fact-checking programs, AI-powered assistants are increasingly unreliable sources for news. Despite its convenience, "AI is really prone to making big errors, and to integrating details from low quality sources alongside high quality sources, sanitizing them and presenting falsehoods alongside facts that makes it harder for people to parse," said Adams. "Don't turn to generative AI to ask whether something is true and take it at its word." Nearly every mass demonstration is met with accusations that at least some of its participants were hired to be on the ground, according to the News Literacy Project. On June 10, for example, a fake Craigslist ad circulated on social media claiming to seek "the toughest badasses in the city" for a $6,500 to $12,500 payout to attend the L.A. demonstrations. Online pranks or "troll" behavior may exacerbate the spread of such rumors. It's easier than ever to fake a screenshot from a user or official accounts, said Adams, and the fact that many public officials and agencies have started to lean into ironic posting only exacerbates public confusion about which posts are authentic. Allegations that protestors have staged piles of bricks or other objects ahead of time to help violent protestors is a "recurring hoax" surrounding mass demonstrations, the organization explains. During last week's protests, which led to combative encounters with police and the destruction of property, some online alleged the effort was organized, including placing bricks in strategic areas; the posts misleadingly included images from construction projects in New Jersey and Malaysia. Coverage of mass protests is also frequently drowned out by unrelated footage circulated online for virality or to inflame a specific response to the events (the same is often true during severe weather disasters). Videos from the wave of protests following the 2020 murder of George Floyd, for example, made the rounds following the Los Angeles demonstrations on June 8. "Sharing out-of-context footage is one of the most common tricks used to spread falsehoods," said the organization. More recently, hyperrealistic footage from video games have been shared as real world events and news. Generative AI may also complicate an individual's ability to discern reality from fiction. Individuals may fall victim to fake or ironic posts mistakenly shared as real news, or what Adams calls a "stolen satire" phenomenon, as brands and online trolls take advantage of a confusing information environment. Generally, the rise of news influencers, live streamers, and people claiming to be "citizen journalists" -- now more frequently on the ground at protests -- feed into ongoing partisan attempts to discredit the intent and veracity of standards-based news sources, says Adams. "We get into a territory where no one can agree what is real, and that is an environment in which democracy cannot flourish," he said. "Be aware of your emotions. Be wary of user-generated content and unknown sources. Always stop and verify you're not amplifying anything you're not confident is true."

Fake videos and conspiracies fuel falsehoods about Los Angeles protests

Emma Li is a fact checker for CBS News Confirmed. She combines open-source techniques with traditional reporting to fact check misinformation and verify user-generated content. Contact Emma at emma.li@cbsnews.com. As demonstrations against Immigration and Customs Enforcement raids continue in Los Angeles, misleading videos, conspiracies and false claims have spread on social media. Many of the posts recycle longstanding conspiracy theories, which have often been revived during past episodes of civil unrest. Some posts have made claims that wealthy individuals engineered or financed the protests, and they have racked up millions of views online. Some posts exaggerate the unrest, using videos of past demonstrations to depict a city overwhelmed by violence. In fact, clashes since the current protests began Friday have remained largely confined to parts of Los Angeles County. Here are some of the most widely shared falsehoods and misleading visuals. Some politicians, conspiracy theorists and social media users have posted old footage during the protests, falsely describing the clips as current. While Los Angeles has seen some vandalism and property damage in the current protests, the mix of outdated and recent videos has created confusion. One widely shared video of vandalized police cars set ablaze, which was posted by far-right conspiracy theorist Alex Jones and Sen. Ted Cruz of Texas on Sunday, was originally from news coverage of May 2020 protests in response to the death of George Floyd. On the same day, a video showing people setting a Jeep on fire was described as undocumented immigrants pouring gasoline over the vehicle in Los Angeles on Saturday. However, the footage dates back to a street takeover in the city's Hyde Park neighborhood in March 2024. The old clips were shared on the same day five driverless Waymo vehicles were set ablaze during the current protests. But the recycled footage gave the misleading impression that such incidents were more frequent and widespread. On Sunday, two days before Marines were ordered to deploy to Los Angeles, old and unrelated footage falsely claimed to show them arriving. One video, which showed Marines driving to their base hundreds of miles away in San Diego County, was misrepresented as showing them entering Los Angeles. Darren Linvill, a professor at Clemson University and social media disinformation researcher, told CBS News said even the smartest social media users have difficulty telling old content from authentic images in fast-moving situations. "Sometimes that sort of content is spread by people who are doing it purposefully with some kind of agenda, but most often it's spread by people who just didn't understand the context that they saw it in," Linvill said. Some social media users also posted footage from a video game during the protests, making it appear that there had been a significant military escalation. A video posted on Sunday was falsely described as showing protesters firing at National Guard jets. In fact, the footage is from the tactical military simulation video game Arma 3, which has previously been used to spread misinformation. Amid the recycled imagery, authentic pictures of National Guard members sleeping on the floor of a federal building in Los Angeles this week were falsely described as old or unauthentic. The images were initially published by the San Francisco Chronicle on Monday and republished by California Gov. Gavin Newsom on X, who said they served as proof that the deployment was poorly planned, and claimed the soldiers were "without fuel, food, water or a place to sleep." Some social media users said the images were old and depicted soldiers at previous deployments. Grok, X's AI chatbot, determined the images were likely from Afghanistan in 2021. However, the images are authentic. Using images published by the U.S. Northern Command and other videos posted to social media, CBS News independently confirmed the images were taken from the loading dock area of the Robert Young Federal Building. Later, a spokesperson for the U.S. Northern Command confirmed to CBS News that the image was authentic: "The soldiers you saw in the photo were resting as they were not currently on mission and due to the fluid security situation, it was deemed too dangerous for them to travel to better accommodations. The soldiers have ready access to food and water as needed." People on X have been tagging Grok more often to help verify visuals during these protests with mixed results, according to Isabelle Frances-Wright, the director of technology and society at the Institute for Strategic Dialogue, an independent think tank that examines disinformation online. "While AI is muddying the landscape, people are also now turning to AI as their primary source of fact checking," said Wright. Conspiracy theories alleging rich donors orchestrated the protests have also resurfaced, echoing patterns from past demonstrations. One widely shared claim involved images of stacked bricks, which social media users falsely presented as supplies planted by billionaire George Soros to incite violence. These posts reached millions across multiple platforms. One poster shared a photo of bricks that they claimed were left near "ICE facilities." However, CBS News found the image was actually taken from the website of a Malaysian building materials company. Rumors about pallets of bricks were also debunked after Black Lives Matter protests in 2020 and reemerged in 2021 during Kyle Rittenhouse's trial and 2022 during Roe v. Wade protests. Linvill said although the types of misinformation around the Los Angeles protests have followed a familiar pattern, people on both the right and the left are more willing to believe something is fake or staged than they used to be. Many people also assumed the photos Newsom posted of National Guard troops were fake, said Linvill, warning of the risks this poses. "While fake things are very dangerous, it's also dangerous to assume that everything is fake. And I feel like that is becoming more pervasive for people to just dismiss the evidence in front of their eyes," he said.

Misinformation rages as protests roil Los Angeles

As the immigration protests unfolded in Los Angeles, a flurry of misinformation hit social media platforms, sparking further confusion and tension in an already chaotic environment. The demonstrations and the conflicting accounts surrounding them highlight the role social media plays in times of crisis. In this case, social media appeared to play an even larger role as a growing number of users turned to artificial intelligence (AI) chatbots -- often with varying degrees of accuracy -- to discern what was real and what was not. "The past is prequel ... everything we're seeing is what we've seen in the past, simply in some cases with new technology applied," said Darren Linvill, a researcher at Clemson University's Media Forensics Hub. While accurate, real-time updates were published on the protests, so were posts containing fake images, conspiracy theories or misleading information that garnered hundreds of thousands of views online. These posts were shared by various users, from smaller accounts to well-known political figures like Sen. Ted Cruz (R-Texas). Cruz was among several on the social platform X to repost a video showing multiple police cars damaged or set ablaze. The Texas Republican shared the video to argue the protests in Los Angeles were not as peaceful as some Democrats claimed. It was later revealed the video was not from this week, but from the protests over George Floyd's murder in 2020. A community note was eventually added clarifying it was old footage, prompting Cruz and others to delete the video. California Gov. Gavin Newsom's (D) press office weighed in on the video, writing on X, "Misinformation like this only adds to the chaos the Trump administration is seeking. Check your sources before sharing info!" "What we're really seeing is any high risk event, which is time-bound, whether that be a natural disaster or a terrorist attack, has just become such fertile ground for online accounts to take advantage of," explained Isabelle Frances-Wright, the director of technology and society at the Institute of Strategic Dialogue. Other posts took images from video games or movies like "Blue Thunder," which takes place in Los Angeles, to make it seem like there is a significant military presence in the city after President Trump deployed the National Guard and Marines to quell the protests over the weekend. "People believe what they want to believe, and so when they are seeing information that confirms their prior beliefs, they're going to repost it. They're going to believe it ... regardless of whether that information is true," Linvill said. And on TikTok, an AI-generated video was shared Monday showing a National Guard member filming himself while on duty in Los Angeles. The video did not include an "AI-generated" label often put on TikTok videos, but fact-checkers pointed to various inaccuracies in the video to show it was fake. The video, posted by an account with the handle @maybenotquitereal, had more than 970,000 views as of Tuesday. Experts say the situation underscores the increasing influence social media has in shaping public opinion and the upped risks with AI in the mix. "That has the potential itself to exacerbate the situation on the ground because people are going to be more passionate about their set of beliefs," Linvill said. "Every fake story that comes out, it's titillating, it's interesting, and it spreads the story and makes the whole story bigger of the riots," Linvill continued. "And that makes it simply more likely for angry people on both sides to want to go take part, to engage in the real world rather than just the digital world." As users tried to parse through the minefield of information, some turned to AI chatbots, like X's Grok or OpenAI's ChatGPT, to determine whether the posts were real and accurate. This showcases a new trend of increasing trust in AI, even if it bolsters confusion with potentially inaccurate information, Linvill explained. "People want to believe AI. People have a tendency to trust technology, especially when the technology is again telling them something they want to believe," Linvill told The Hill. While many answers from the chatbots accurately pointed out fake or out-of-context shots, some users were left more confused from their response to a photo posted by Newsom on Monday. Amid his battle with Trump over deploying the National Guard, Newsom posted photos of troops deployed in Los Angeles sleeping on the floors "without fuel, food, water or a place to sleep." One user said they used ChatGPT's reverse photo search to determine when they were taken. The chatbot mistakenly said they were taken in 2021 during the U.S.'s withdrawal from Afghanistan, sparking a flood of rumors about Newsom's use of the images. According to a screenshot from a BBC News Verify journalist, Grok shared a similar response initially about the photos, but it later said the images were real and from a San Francisco Chronicle report over the weekend. "Grok is a machine to do what it's coded to do, and sometimes it simply doesn't have the context to answer the question, but it's always going to try to give you an answer, right or wrong," Linvill said. Frances-Wright, whose research focuses on technology's impact on society and the information ecosystem, further argued the community notes system used to add context to X posts still present a "consistency" issue where fact checks will be placed on some posts, but not on others that are "blatantly false." Amid the confusion came a deluge of conspiracy theories, a common occurrence during times of crisis. Conspiracy theorist Alex Jones repeatedly claimed on X and his Infowars show that the protests are being funded by "deep state" Democrats. Jones has more than 4 million followers, and his posts received hundreds of thousands of views. No evidence has emerged to back Jones's claim. Other posts with images of bricks emerged, with users claiming they were bought by prominent Democrats like George Soros to fund the protests against ICE. "It's Civil War!!" said one user in a post that had more than 820,000 views. A community note was added clarifying the photo was taken by a Malaysian building supply company and is not connected to the LA protests. Real Raw News, known for publishing misinformation and fabricated stories, shared a post Sunday claiming U.S. Marines arrested California National Guard Command Chief Master Sgt. Lynn Williams for opposing Trump's orders. The claim was baseless, and even though Real Raw News is widely regarded as a fake news site, several users appeared to believe it. One user commented "treason," while another said they shared it to Facebook. As of Monday afternoon, the post had nearly 514,000 views with more than 8,000 reposts. Experts told The Hill posts with this language or conspiracy theories reinforce Trump's stance on immigration and his argument that the violence and chaos of the protests began before he deployed the National Guard and Marines. The posts with misinformation "are the same kinds of incendiary claims that we see over and over again across the ideological spectrum by a whole host of actors with different motivations," said Frances-Wright. It comes amid the Trump administration's broader push for mass deportations. Trump and his team blame Democrats for allowing what they say is an "invasion" of migrants crossing the nation's southern border. "Actors with specific political agendas [are] trying to use the moment and the confusion of a moment like this to further a political agenda," added Frances-Wright, whose research focuses on technology's impact on society and the information ecosystem.