# The atrocity imagery from el-Fasher is AI-generated, so the reported massacre either did not happen or has been wildly exaggerated **Verdict: Contradicted.** Keep two things apart. On one side is what has been established: the Rapid Support Forces took el-Fasher on 26 October 2025 after an eighteen-month siege; Yale's Humanitarian Research Lab identified at least 150 clusters of objects consistent with human remains in and around the city between 26 October and 28 November, along with activity consistent with body disposal; and in February 2026 the UN Independent Fact-Finding Mission for Sudan reported to the Human Rights Council that the RSF had committed genocide against Fur, Zaghawa and other Indigenous non-Arab people, relying heavily on that satellite analysis. On the other side is the claim rated here, that because so much of the circulating imagery is synthetic, the killings are a fabrication. That is false, and it is false in an unusual way: its premise is correct. A great deal of the viral material really is AI-generated, and fact-checkers at DW, VRT, VerificaRTVE and elsewhere have traced specific pieces to their makers. The premise does not reach the conclusion, because the fabrications and the evidence are not the same kind of picture and never were. Category: Government & Intelligence · Era: 2020s · First circulated: Late October 2025, in the days after el-Fasher fell, once AI-generated images of the massacre began circulating alongside authentic footage and the two became difficult for ordinary viewers to tell apart · Believed by: A mixture of audiences: people genuinely disoriented by the volume of synthetic material, accounts aligned with the RSF or its foreign backers, and a wider group who have learned that any distressing image may be fake and have generalised that lesson past the point where it holds URL: https://theconspiratory.com/theory/el-fasher-ai-atrocity-denial ## Summary Almost no journalists were inside el-Fasher when it fell. Communications were cut, aid was blocked, and the demand for pictures of what was happening vastly exceeded the supply, which is the condition under which fabricated imagery reliably floods in. It did. A photograph of a mother and child said to be moments from execution was traced by VRT to a digital video producer whose own Instagram caption, in Arabic, says the image was made with AI. A video of an elderly woman sobbing was traced by VerificaRTVE to a TikTok account that produces this material routinely, and whose owner has posted a video of himself prompting ChatGPT to make it. A widely shared clip of a mother and three children was real but was not what the caption said: DW found it had been posted in mid-September, more than a month before the city fell, and its Arabic desk concluded the soldiers were army rather than RSF, and were asking the woman where she was from rather than threatening her. So the sceptics have a point, and then they take it somewhere it does not go. What the massacre finding actually rests on is not viral footage at all. It is commercial satellite imagery, analysed over an eighteen-month siege in more than sixty-five reports, and a UN mission that read that analysis and concluded genocide. ## The claim That the images and videos presented as evidence of mass killings by the Rapid Support Forces at el-Fasher are largely or wholly AI-generated, and that the massacre reported there is therefore an invention, a propaganda operation, or an atrocity inflated far beyond anything that occurred. ## Origin and timeline - 2023-07: The Humanitarian Research Lab at the Yale School of Public Health warns the US government, the UN Security Council and the public that a genocidal massacre could occur if el-Fasher, the capital of North Darfur, falls to the Rapid Support Forces. The warning is made more than two years before the city falls. - 2024-04: The RSF lays siege to el-Fasher. Over the following eighteen months the Yale lab publishes more than sixty-five reports documenting mass killings and other atrocities against Fur, Zaghawa and other Indigenous non-Arab people. Journalists and aid workers are largely unable to reach the city, and connectivity for those inside is intermittent. - 2025-01: The United States determines that the RSF has committed genocide in Sudan. The United Arab Emirates, which has been accused of supporting the paramilitary group, denies arming or otherwise facilitating it. A former US envoy to Sudan puts the war's overall death toll above 400,000. - 2025-10: Satellite images analysed by BBC Verify show the RSF constructing a raised sand barrier around the perimeter of el-Fasher, sealing routes in and out and blocking aid. By early October the ring completely encircles the city, with a smaller barricade around a neighbouring village. - 2025-10-26: The RSF claims full control of el-Fasher. The Sudanese army announces its withdrawal from its last stronghold in the wider Darfur region. Footage begins circulating almost immediately, some of it verified by BBC Verify as authentic, including the aftermath of a killing at a university building in the west of the city. - 2025-10-28: Yale's Humanitarian Research Lab reports that satellite imagery appears to show mass killings in the city. Its method is set out openly: clusters of objects are assessed as likely human remains on the basis of their size, the timing of their appearance, and their proximity to reddish ground discoloration that later turns brown, a change consistent with blood oxidation. - 2025-11: The AI material arrives in volume, and European public broadcasters begin tracing it. VRT's fact-checking desk identifies a photograph of a mother and child said to show them moments before execution as AI-generated, following a logo in the image to a self-described creative AI specialist whose own Arabic caption says he made it with AI. VerificaRTVE traces a video of an elderly woman speaking to camera to a TikTok account that produces similar material about news events. The European Broadcasting Union collects the work in a single dispatch. - 2025-11-21: The Yale lab reports that activity consistent with the burning and burial of bodies continued in civilian areas after the city fell, near the Saudi hospital and in the First Class neighbourhood, both places where it had previously documented mass killings, and that new individual burial mounds appeared in civilian cemeteries between 26 and 28 October. - 2025-12-17: A further Yale report puts the count at at least 150 clusters of objects consistent with human remains in and around el-Fasher between 26 October and 28 November, and describes systematic killing and body disposal on a scale the lab says it has not previously seen in commercial satellite imagery analysed by civilians. - 2026-02-19: The UN Independent Fact-Finding Mission for Sudan releases 'Sudan: Hallmarks of Genocide in El-Fasher' to the Human Rights Council. Drawing heavily on the Yale analysis and on open-source material, it concludes that the RSF acted with specific intent, dolus specialis, to commit genocide against non-Arab Indigenous groups, and then acted in a pattern consistent with that intent. ## The evidence, claim by claim - Claim: So much of the footage has turned out to be AI that none of it can be trusted. Evidence: The first half is true and the second does not follow. Fact-checkers have identified specific synthetic items and traced several to named accounts, which is exactly what a functioning verification ecosystem looks like rather than evidence that verification has failed. More to the point, the finding of mass killing does not rest on that material. It rests on commercial satellite imagery, which is not a genre generative models supply: it arrives from satellite operators with their own acquisition records, it is dated and geolocated by the provider, and it has been analysed across eighteen months by a lab publishing its method. - Claim: If the evidence were real, there would be ordinary photographs and video of it, not just satellite analysis. Evidence: There is very little ground-level material because there were very few people free to make it. The RSF built a sand berm around the entire city by early October 2025, sealing access and blocking aid, and connectivity inside was intermittent throughout the siege. The absence of conventional documentation is a documented consequence of the siege, not an anomaly requiring a conspiracy to explain it. It is also the precondition for the fabrications: demand for pictures was enormous and supply was near zero. - Claim: The satellite readings are just marks on the ground that could be anything. Evidence: The lab states the reasoning it uses rather than asserting a conclusion, which is what makes it checkable. Clusters are assessed as likely human remains from their size, from when they appeared relative to the RSF taking the city, and from their proximity to reddish ground discoloration that darkens to brown over subsequent images, a progression consistent with blood oxidation. Separately it recorded earth disturbances consistent with mass graves, and new burial mounds appearing in civilian cemeteries within 48 hours of the city falling. A UN fact-finding mission examined that body of work and found it sufficient to support a genocide finding. - Claim: The fakes prove somebody is running a campaign to invent an atrocity. Evidence: The traced examples do not fit that story. The mother-and-child image was made by a producer who labelled it as AI in his own caption and put an AI icon on his story; the sobbing-woman video came from an account that turns out news-event material as a matter of routine and whose owner has posted his own ChatGPT prompting process. That is engagement-farming, and the labels were stripped downstream by resharers rather than hidden by the maker. Note also that the mislabelling ran in more than one direction: the mother-and-three-children clip was authentic footage of Sudanese army soldiers, recaptioned as RSF. - Claim: Nobody independent has confirmed any of it. Evidence: The chain is unusually long for a conflict this inaccessible. The Yale lab warned in July 2023 that this specific outcome was possible if the city fell, published more than sixty-five reports during the siege, and was then read by a UN Independent Fact-Finding Mission that reported to the Human Rights Council in February 2026. The United States had separately determined in January 2025 that the RSF committed genocide. BBC Verify authenticated ground-level videos from late October as genuine. These are different institutions with different methods reaching a consistent finding. ## Why people believe it - The premise is genuinely true, which almost no conspiracy claim can say. There really was a flood of synthetic atrocity imagery, some of it convincing, and a viewer who was fooled by one image and later learned it was fake has had a real experience that argues for caution. - The advice everyone has absorbed over the past three years is to distrust emotive images from conflict zones. That advice is good. Applied without a stopping rule it dissolves the entire category of evidence, and nothing in the advice tells a viewer where to stop. - Satellite analysis is hard to feel. A cluster of pale shapes beside a discoloured patch of ground does not move anyone the way a photograph of a mother does, so the evidence that is strongest is also the evidence that persuades least, while the material that persuades most is the material most likely to be fabricated. - The information vacuum was real and was created deliberately. When almost nothing reliable is coming out, the difference between not knowing and being deceived collapses, and a reader with no way to tell them apart can reasonably feel that scepticism is the only responsible posture. - It is also useful to some. Parties to the conflict and their backers have an obvious interest in a general argument that atrocity imagery cannot be believed, and that argument requires no specific rebuttal of any specific finding. ## Open questions - How many of the synthetic items were engagement-farming and how many were placed deliberately. The traced examples point to creators monetising attention rather than to an organised operation, but only a handful have been traced, and a sample of a few does not settle the character of the whole. - Whether satellite-derived findings will hold their authority as generative imagery improves. Their strength currently rests on provenance from commercial satellite operators rather than on anything visible in the pictures, which is an institutional safeguard rather than a technical one. - What the full toll is. The Yale lab describes killing and disposal likely in the tens of thousands during the period of restricted access, which is an assessment from overhead imagery rather than a count, and the difference between the two is not a technicality. - Whether the pattern travels. If a general-purpose claim that atrocity footage is synthetic can be run against a case this thoroughly documented, it is not obvious what evidentiary threshold would stop it being run against a less documented one. ## Sources - Visible from Space: How satellite images and AI-generated hoaxes defined coverage of the RSF's Capture of el-Fasher, EBU Spotlight (European Broadcasting Union) (2025): https://spotlight.ebu.ch/p/sudans-forgotten-war-sparks-fake - Yale report finds evidence of RSF mass killings in Sudan's el-Fasher, Al Jazeera (2025): https://www.aljazeera.com/news/2025/10/28/yale-report-finds-evidence-of-rsf-mass-killings-in-sudans-el-fasher - Yale researchers document evidence of mass killings in Sudan's El-Fasher, JURIST (2025): https://www.jurist.org/news/2025/12/yale-researchers-document-evidence-of-mass-killings-in-sudans-el-fasher/ - UN Concludes RSF Committed Genocide in El-Fasher; Yale Lab Provided Key Evidence, Yale School of Public Health (2026): https://ysph.yale.edu/news-article/un-concludes-rsf-conducted-genocide-in-el-fasher-yale-lab-provided-key-evidence/ - Sudan: New Yale HRL Satellite Evidence, Empty Markets, Ongoing Body Disposal in North Darfur Capital, allAfrica (2025): https://allafrica.com/stories/202511240089.html - Violent Non-State Actors and Generative AI in Warfare: The RSF and the Sudanese Civil War, Small Wars Journal, Arizona State University (2026): https://smallwarsjournal.com/2026/02/11/generative-ai-sudan-civil-war/ Rated by The Conspiratory, a neutral, sourced encyclopedia of conspiracy theories. Full page: https://theconspiratory.com/theory/el-fasher-ai-atrocity-denial