A mob admits he was “paid to be here.” National Guard soldiers film themselves being attacked by “balloons full of oil.” A young man declared his intention to “protest peacefully” before throwing Molotov cocktails.
These are a few things do not have It's happening on the streets of Los Angeles this week. But if you're receiving your news from AI, you might think they are.
The fake AI-generated videos, photos and factoids about the ongoing protests in LA are spreading like wildfires across social media, particularly on Elon Musk's Anything Goes Social Network X (formerly Twitter).
These completely synthetic chunks of rage, made using freely available video and image generation software, usually confirm existing stories about protests. For example, the unfounded idea that it is secretly funded and equipped by mysterious outside factions.
And while some are labelled technically as parody, many users have missed these disclaimers, assuming that the realistic footage is a real documentary of an event on the ground.
“Hey guys! Bob in National Guard duties. Stack, I'll look behind the scenes and see how to prepare crowd-control gear for today's gasting,” says the soldier simulated in a video of the virus Tiktok, exposed by France 24.
“Hey team!” he follows immediately. “Bob here, this is insane! They're chucking us balloons full of oil, look!”
Another fake video posted on X features a male influencer wearing a too thick, clean t-shirt from the riot. “Why are you rioting?” he asks the masked man. “I don't know, I was paid to be here, and I just want to destroy things,” the man replies.
AI tries to fact check AI
Meanwhile, chatbots like Openai's ChatGpt and X's built-in Grok have incorrectly answered user questions about events in the City of Angels.
According to CBS News, both Grok and Chatgpt mistakenly claimed that a photo of a National Guard member was sleeping with a “probably born” in Afghanistan in 2021.
It also reportedly claimed that a virus photo of a brick stacked on a pallet (which was promoted by a right-wing disinformation merchant as evidence of external funds) was “probably” taken from the LA protest. In fact, the photos showed random streets in New Jersey, even when they were informed of the truth stuck on a gun.
The fakes generated by AI are just one of the new instruments in aging disinformation orchestras. Photos recycled from past protests and events, out-of-context photos, and footage of disguised video games – not among the Peddlers of partisan rage since 2020, but are widely shared, including Republican Sen. Ted Cruz (formed in this respect).
“Photos are easy to manipulate, and the idea was there,” James Cohen, a media professor and expert at Internet literacy at Cuny Queens College, told Politico.
“But when it comes to videos, we have just been trained as individual societies to believe in videos. Until recently, we had no real opportunity to assume that we can forge videos on a scale that is being forged at this point.”
Culture War Ammunition
Most of the videos were seen Independence It was clearly aimed at a conservative audience and was designed to reinforce or refer to right-wing topics. However, at least one Tiktok account has seen over 300,000 views on the video on Wednesday evening, and appears to be aiming to be a progressive interest in stirring up a message of solidarity with immigrants.
Some are obviously jokes. Others joke vaguely. Often there is a tag indicating that the video is being generated by AI. However, in most cases, this important information is easily overlooked.
Unfortunately, this issue could get worse in the future. The Republican flagship, Big Beautiful Building, includes a suspension on all state regulations for AI, preventing state governments from intervening for 10 years.
Videos generated by AI can be difficult to communicate from the real thing, but there are ways. They often have a suspiciously clean and shiny look, like the same groomed space charm as Kendall Jenner's infamous Pepsi protest ads. Those people are often oddly beautiful, like a fusion of millions of magazine photography.
The Better Business Bureau also recommends scrutinizing important details such as fingers and coat buttons, which often makes no sense in a thorough examination. The writing is also frequently blurred and readable. It's a mess of letters and letter-like shapes.
Background numbers can behave strangely or repeatedly, or even in ways that are physically impossible. If you are in doubt, search on Google to see if a trusted media organization or individual journalist has seen or exposed what you are seeing.
Above all, keep an eye on what appears to be a complete confirmation of existing beliefs. That might be too good.
