# Viral videos showing dramatic ICE raids at Walmart, McDonald's, and other workplaces are authentic footage of real immigration enforcement **Verdict: Contradicted.** Narrowly scoped, this is settled. The specific clips that went viral in 2026, the Walmart line-up, the McDonald's raid, the food-delivery-rider ambush, the classroom and Pizza Hut scenes, are AI-generated fabrications, not footage of real events. They were mass-produced with text-to-video tools, principally OpenAI's Sora, by engagement accounts such as 'USA Journey 897,' and were identified as synthetic by Full Fact, HuffPost, Euronews, Snopes, AFP, and Reuters through Sora watermarks, uniform 10- and 15-second run times, and telltale errors (misspelled badges reading 'POICE,' vans labeled 'IMMIGRATION AND CERS,' impossible physics). This file rates only the authenticity of those viral clips; it takes no position on real immigration enforcement, which is a separate matter documented by conventional reporting. The harder problem the fakes create, that genuine footage can now be waved away as 'just AI,' is the liar's dividend, addressed below. Category: Science, Space & Technology · Era: 2020s · First circulated: Circulated from late 2025 into 2026, after OpenAI's Sora 2 became widely accessible; the account 'USA Journey 897' and similar pages were reported through winter and spring 2026 · Believed by: A large casual-scrolling audience on Facebook, Instagram, and YouTube rather than a defined movement; individual clips drew millions of views and comment sections show many viewers treating them as real, while others correctly flag them as AI URL: https://theconspiratory.com/theory/ai-ice-raid-videos ## Summary Through late 2025 and into 2026, hundreds of AI-generated videos depicting supposed ICE immigration raids, at Walmarts, McDonald's, workplaces, schools, and on city streets, racked up millions of views across Facebook, Instagram, and YouTube. Engagement accounts such as 'USA Journey 897' mass-produced them with text-to-video tools, principally OpenAI's Sora. Fact-checkers including Full Fact, HuffPost, Euronews, and Snopes confirmed the specific viral clips are synthetic fabrications, not footage of real events, identifying them through leftover Sora watermarks, uniform clip lengths, and visual errors. This case file debunks the authenticity of those clips only. It takes no position on actual immigration enforcement, which is documented separately by conventional journalism. The deeper story is the 'liar's dividend': a flood of convincing fakes does double damage, spreading false events while also giving anyone cover to dismiss genuine footage as 'just AI.' ## The claim That widely shared videos showing ICE agents carrying out dramatic workplace raids, lining up Walmart and McDonald's employees, ambushing delivery riders, storming classrooms, are authentic recordings of real immigration-enforcement operations captured by bystanders, rather than computer-generated fabrications. ## Origin and timeline - 2018: Legal scholars Bobby Chesney and Danielle Citron publish work naming the 'liar's dividend': as the public learns that convincing fakes exist, bad actors gain the ability to dismiss authentic recordings as fabricated. The concept becomes the frame for understanding the later ICE-video wave. - 2024–2025: Consumer text-to-video generators reach a quality where short clips of crowds, uniforms, and street scenes look plausible at a glance and at phone-screen size, lowering the skill and cost needed to stage a convincing 'news' moment. - 2025-10: OpenAI's Sora 2 becomes widely accessible, letting anyone generate short, realistic video from a text prompt. Its output carries a visible corner watermark and embedded C2PA provenance metadata, and clips default to fixed lengths, details that later help identify the fakes. - 2025–2026: Synthetic clips staged to look like ICE raids begin spreading on Facebook, Instagram, TikTok, and YouTube, mixing in among authentic bystander footage of real enforcement and making it harder for viewers to tell what actually happened. - 2026-02: German public broadcaster ZDF recalls a US correspondent after an AI-generated clip, showing children clinging to their mother as she is arrested, appears in one of its reports; the video carried a Sora watermark. The episode shows the fakes fooling not just casual scrollers but a national newsroom. - 2026: Independent researcher Chad Loder flags the Facebook account 'USA Journey 897' to 404 Media. Its Sora-generated raid videos, styled to resemble official ICE and DHS social posts, draw enormous reach; one clip is reported at roughly 40 million views and a Walmart clip at more than 4 million. - 2026: Reporting by The Cube (Euronews) counts more than 200 short AI clips portraying ICE agents in staged scenarios (chased by teachers, brawling in bars, arrested by the NYPD), many without clear labeling, collectively drawing millions of views. - 2026: After scrutiny, the operator of 'USA Journey 897' adds a disclaimer describing the videos as created with AI 'for entertainment and creative purposes only,' an acknowledgment that arrives long after the clips have already spread as apparent real events. - 2026-07: A fabricated screenshot of a Truth Social post, falsely attributed to President Trump and proposing to use ICE facilities to detain political opponents, spreads alongside the video wave. Snopes and Lead Stories trace it to a satirical account and rate it fake, illustrating how the synthetic flood spans video and text. ## The evidence, claim by claim - Claim: The viral clips are real bystander footage of ICE operations at Walmart, McDonald's, and other workplaces. Evidence: They are not. Multiple fact-checking organizations, Full Fact, HuffPost, Euronews, and Snopes among them, examined the specific viral clips and concluded they are AI-generated. The videos carry the hallmarks of text-to-video output rather than a camera: identical short run times, corner watermarks, and physically impossible details. The question is not close; it is a media-forensics finding, not a matter of opinion. - Claim: Nothing in the footage marks it as fake; it looks like an ordinary phone recording. Evidence: On the contrary, the clips are riddled with tells. Reported examples include a van labeled 'IMMIGRATION AND CERS,' an agent's badge reading 'POICE,' a worker doing an unnatural sideways walk, garbled background signage, misspelled uniform logos, and steam rising through a solid surface. These are the characteristic artifacts of current generative video, not the errors a real scene produces. - Claim: The clips came from on-the-scene sources, not a content operation. Evidence: They were traced to engagement accounts, most prominently 'USA Journey 897,' flagged to 404 Media by researcher Chad Loder, that mass-produced dozens of similar scenes. Euronews's investigative strand counted more than 200 such clips. The uniformity of format and volume of output point to automated generation by a small number of accounts, not many independent witnesses. - Claim: The tool used cannot be identified, so the 'AI' label is guesswork. Evidence: The tool is identifiable. Clips ran exactly 10 or 15 seconds, matching Sora's default output, and bore Sora's visible watermark, which some posters tried to obscure by superimposing text over the three corners where the mark appears. Sora files also embed C2PA provenance metadata that verification tools can surface. The attribution to OpenAI's Sora rests on concrete signatures, not inference. - Claim: The account labeled the videos as AI, so viewers always knew they were fictional. Evidence: The disclaimer came late and did little. The operator of 'USA Journey 897' added an 'entertainment and creative purposes only' note only after the clips had already gone viral as apparent real events, and comment sections show many viewers reacting as though the raids were genuine. A label buried on a page does not travel with a clip that is screen-recorded, re-uploaded, and reshared across platforms. - Claim: Because these clips are fake, footage of real ICE enforcement must be fake too. Evidence: This does not follow, and it is precisely the trap. Debunking these specific fabrications says nothing about the authenticity of any other, unrelated recording. Genuine enforcement footage is documented through conventional reporting and is assessed on its own evidence. Treating 'some clips are AI' as 'therefore all clips are AI' is the liar's dividend at work, and this file explicitly declines to make that leap. ## Why people believe it - The clips are short, low-resolution, and viewed on a phone, exactly the conditions under which generative-video artifacts are hardest to notice; a real bystander video would look much the same at a glance, so the format itself lends the fakes credibility. - They were engineered to mimic the look of official ICE and DHS social-media posts, borrowing the visual grammar of a genuine government account, which primes viewers to read them as authentic rather than staged. - They dramatize a subject people are already anxious and polarized about, so the clips confirm what many viewers half-expect to see; content that fits a prior belief is shared quickly and scrutinized slowly. - Recommendation algorithms reward watch time and outrage, and vivid fakes can out-perform mundane real footage, so the synthetic clips are actively surfaced to large audiences while measured corrections are not. - Recognizing AI video takes a specific skill set (knowing to check clip length, watermarks, hands, text, and physics) that most viewers have not been taught, so an ordinary person has no easy way to tell a Sora clip from a phone recording. ## Open questions - How much of the total reach was driven by a handful of high-volume accounts versus organic resharing is not fully mapped; the counts published so far (roughly 200 clips, tens of millions of views on individual posts) are lower bounds from specific investigations, not a complete census. - Whether platform monetization actively rewarded these pages, and how much their operators earned, is only partly documented; reporting has raised the question of ad revenue flowing to dehumanizing synthetic content, but the full financial picture is not public. - Whether provenance standards like C2PA and visible watermarks will meaningfully help, given that watermarks can be cropped or covered and metadata stripped on re-upload, is an open and fast-moving question rather than a solved one. - How durable the labeling and detection response will be as generation quality improves and default tells (fixed clip lengths, visible watermarks) are engineered away is genuinely unresolved; today's reliable signatures may not survive the next model generation. ## Latest developments - 2026-07-18T15:00Z: As of mid-July 2026 the pattern is still active and evolving. Fact-checkers continue to log new Sora-made ICE scenarios (classroom raids, NYPD-versus-ICE confrontations, delivery-rider ambushes), and a fabricated Trump 'Truth Social' post about ICE detention facilities spread in the same window, showing the synthetic flood now spans both video and text. The specific clips remain debunked as AI; the broader liar's-dividend problem they feed is unresolved. ## Sources - AI-Generated Sora Videos of ICE Raids Are Wildly Viral on Facebook, 404 Media (2026): https://www.404media.co/ai-generated-videos-of-ice-raids-are-wildly-viral-on-facebook/ - Flood of AI-generated ICE videos risks undermining trust in real footage, experts warn, Euronews (The Cube) (2026): https://www.euronews.com/my-europe/2026/03/02/flood-of-ai-generated-ice-videos-risks-undermining-trust-in-real-footage-experts-warn - These AI ICE Videos Are All Over Facebook. The Fact That So Many Are Falling For Them Should Scare You., HuffPost (2026): https://www.huffpost.com/entry/viral-ai-ice-raid-videos_l_69163baee4b0df3e16be4163 - Videos of people giving their views on immigration are AI fakes, Full Fact (2026): https://fullfact.org/immigration/fake-ai-clips-immigration-vox-pops/ - 20 claims about videos and images related to ICE, investigated, Snopes (2026): https://www.snopes.com/collections/ice-videos-images-claims/ - Video claiming to show ICE agents deported Pizza Hut employees isn't what it seems, Snopes (2026): https://www.snopes.com/fact-check/pizza-hut-ice-video/ - Fact check: Are ICE fakes trying to drown out real videos?, Philippine Daily Inquirer (2026): https://globalnation.inquirer.net/307632/fact-check-are-ice-fakes-trying-to-drown-out-real-videos - Did Trump say ICE facilities should be used to detain communists, Democrats?, Snopes (2026): https://www.snopes.com/fact-check/trump-ice-communists-democrats/ - AI-altered photos and videos of Minneapolis shootings blur reality, NBC News (2026): https://www.nbcnews.com/tech/tech-news/ai-altered-photos-videos-minneapolis-shootings-blur-reality-rcna256552 - Liar's dividend, Wikipedia: https://en.wikipedia.org/wiki/Liar's_dividend Rated by The Conspiratory, a neutral, sourced encyclopedia of conspiracy theories. Full page: https://theconspiratory.com/theory/ai-ice-raid-videos