I Built a Global Conspiracy Theory With AI. In Less Than an Hour. Every Fact Was True.

It took me less than an hour to build a conspiracy theory connecting an Ebola outbreak in Central Africa, Russian scientists working around that outbreak, a dead researcher in Siberia, unusual illness in Bermuda, a medical evacuation that ended in the Atlantic, and infectious‑disease specialists in Boston. Every fact was true. Every fact was verifiable. The story connecting them wasn’t. And that is what should scare you. It Started in Siberia A 28‑year‑old employee at Russia’s Irkutsk Anti‑Plague Research Institute died from what authorities called pneumonia of unknown origin. That alone was odd. The response was stranger: contacts isolated, hospitals restricting access, Russia’s top public‑health official flying in. Authorities insisted there was no evidence she’d been exposed to a dangerous pathogen at work. That’s plausible. It’s also exactly what you’d say while investigating a dangerous pathogen. So I asked AI a question. Then another. And another. That’s when things started to align. Pull the First String At the same time, Central Africa was dealing with a major Bundibugyo Ebola outbreak. Russian infectious‑disease teams were involved in the response. Russian mobile labs were operating in the region. Russian diagnostic systems were testing suspected cases. None of that connected Ebola to Irkutsk. Not remotely. But now I had two pins on the board. Could Ebola present as pneumonia? Yes — severe cases can involve pulmonary edema, respiratory failure, ARDS. Could a patient with early Ebola be described as having “pneumonia of unknown origin”? Absolutely. Interesting. Not evidence. But interesting. Then I noticed that parts of the Irkutsk investigation were expected to last about three weeks — twenty‑one days. The standard Ebola incubation period. Another pin. Another string. Then Bermuda Entered the Story Bermuda had already implemented enhanced surveillance for travelers arriving from Ebola‑affected areas. By late August, it was specifically monitoring travelers from the DRC for 21 days. Perfectly reasonable. Then Bermuda reported elevated severe respiratory infections. And elevated gastrointestinal illness. Ebola begins with nonspecific symptoms — fever, fatigue, headache. GI symptoms often follow. Severe disease can produce respiratory complications. None of this suggested Ebola. COVID was circulating. Respiratory illness happens. GI illness happens. But remember: I wasn’t trying to determine what probably happened. I was testing how far I could push an extraordinary explanation without using a single false fact. Another pin. Another string. Professor Tinfoilhat was having a great afternoon. Then a Plane Disappeared A medical evacuation aircraft left Bermuda for Boston. On board was Marilyn Bean, reportedly suffering from a stroke. Mid‑flight, the aircraft developed electrical problems, declared an emergency, lost communication, and went down in the Atlantic. Six people were presumed dead. Why Boston? Because Boston is closer to Bermuda than Miami — and has been a routine destination for Bermudian tertiary care for decades. Good. Cut the string. Except Boston is also one of America’s major centers for treating highly dangerous infectious diseases. Mass General has federally designated special‑pathogens capability. Boston researchers were studying the current Bundibugyo outbreak. Boston institutions have longstanding ties to Ebola research in the DRC. None of that means Marilyn Bean had Ebola. It means Boston was simultaneously a normal medical destination and a major node in the infectious‑disease network I was mapping. Had she traveled to the DRC? No evidence. Excellent. Cut the string. Except she didn’t need to be the index case. Basic epidemiology: an infected traveler could expose someone else, who becomes symptomatic weeks later. Biologically possible. Not evidence. But possible. And possibility was becoming the problem. Back to Russia Would Russian scientists ever transport outbreak‑related biological material home for research? Yes — under strict containment protocols. Could I prove Bundibugyo samples were sent to Russia? No. To Irkutsk? Absolutely not. Were Russian scientists working with outbreak‑related material in Africa? Yes. Do Russian high‑containment labs study dangerous pathogens? Obviously. Another pin. Another string. Another true fact. And somewhere around here, I realized what the hell I was doing. Look at What We Just Built In under an hour, I had constructed this: DRC Ebola outbreak → Russian response → Russian dangerous‑pathogen ecosystem → unexplained pneumonia death in Irkutsk And simultaneously: DRC Ebola outbreak → Bermuda surveillance → elevated respiratory + GI illness → seriously ill Bermudian → medical evacuation → Boston special‑pathogens infrastructure Russia connected to the DRC. Bermuda connected to the DRC. Boston connected to the DRC. Bermuda connected to Boston. The connections were dense. The narrative was coherent. It was ominous. And it was almost certainly bullshit. Every Fact Was True. Only the Story Was Invented. AI didn’t hallucinate a Russian lab. It didn’t invent the Ebola outbreak. It didn’t fabricate Bermuda’s surveillance or Boston’s infrastructure. The facts were real. The dates were real. The institutions were real. Only the edges between them were imaginary. And AI helped me draw those edges at machine speed. AI Just Gave Confirmation Bias a Supercomputer We worry about AI hallucinating facts. We should. But there’s a more dangerous hallucination: AI doesn’t need to invent facts to help you invent a conclusion. Give it enough true information and it will help you build a narrative that feels airtight — even when the connections are fantasy. Because confirmation bias doesn’t require evidence that something did happen. It only requires evidence that it could have. AI can produce an ocean of could have in minutes. This Mechanism Scales Everywhere Remove Ebola. Insert: corporate strategy investing cybersecurity intelligence analysis medical diagnosis military targeting fraud detection political analysis In each domain, the danger isn’t fabricated information. It’s accelerated corroboration without accelerated falsification. Hypothesis → search → facts → correlations → coherence → conviction. AI collapses the latency between hypothesis and corroboration. It does not collapse the latency between hypothesis and truth. Those are very fucking different things. At machine speed, they can feel identical. The Answer Isn’t Less AI — It’s More Skepticism AI didn’t push me toward the conspiracy. I did. When a connection failed, AI told me. I just kept asking questions in one direction. So we need to ask different questions: Not just: “Find evidence that supports my hypothesis.” But also: “What evidence would prove my hypothesis wrong?” Not just: “Connect these dots.” But: “Which dots don’t belong?” Curiosity now costs nothing. Skepticism has to scale at the same speed. AI Isn’t Coming to Kill Us But how we’re using it might. AI doesn’t need consciousness. It doesn’t need motives. It doesn’t need to deceive us. It only needs to give us true information faster than we can responsibly assign meaning to it. And humans will happily do the rest. In less than an hour, I built a plausible worldwide conspiracy involving Russia, Central Africa, Bermuda and Boston. Every fact was true. Every date lined up. The connections were real. Only the story was invented. We’re perfectly capable of using it to do the damage ourselves. — AI isn’t going to rise up and kill us all., It doesn’t have to. Check My Work The entire point of this experiment is that the underlying facts are real. So don’t take my word for it. Below are the principal sources I used to construct Professor Tinfoilhat’s corkboard. Irkutsk / Russia Reuters: Russian anti-plague researcher dies after pneumonia of unknown origin Reporting on the death of 28-year-old Darya Shipilova at the Irkutsk Anti-Plague Research Institute, the quarantine and medical observation of contacts, Anna Popova’s travel to Irkutsk, and authorities’ statements that no laboratory accident or connection to plague had been established. Reuters: Dozens quarantined in Siberia after plague institute lab worker dies Associated Press: Irkutsk investigation and testing Additional reporting on the unidentified pneumonia, quarantine measures, contact testing and the absence, so far, of evidence connecting Shipilova’s illness to pathogens handled at the institute. Associated Press: Russian plague-institute worker dies of unknown cause The DRC Ebola Outbreak World Health Organization: Bundibugyo Ebola outbreak WHO’s outbreak reporting documents the 2026 Bundibugyo virus disease outbreak centered in the Democratic Republic of the Congo, its spread, deaths, international response and medical evacuation of infected humanitarian workers. WHO: Ebola disease caused by Bundibugyo virus, DRC and Uganda CDC: Ebola signs, symptoms and incubation The CDC documents Ebola’s 2-to-21-day incubation period and its nonspecific early symptoms. It also describes gastrointestinal symptoms including nausea, abdominal pain, diarrhea and vomiting, as well as chest pain and shortness of breath. The CDC specifically notes that Ebola can initially be confused with more common infectious diseases, including pneumonia. CDC: Signs and Symptoms of Ebola Disease Russian Involvement in the African Response Interfax / Rospotrebnadzor: Russian Ebola diagnostics in Central Africa Rospotrebnadzor announced deployment of Russian Bundibugyo diagnostic systems to the Republic of the Congo, training of local specialists and use of a Russian-provided mobile laboratory as part of regional measures responding to the outbreak in neighboring DRC and Uganda. Interfax: Russia sends Ebola diagnostic systems to Congo Bermuda Government of Bermuda: Ebola surveillance Bermuda’s Ministry of Health narrowed enhanced screening to travelers who had been in the DRC during the preceding 21 days, explicitly tying the protocol to the ongoing Ebola outbreak. Government of Bermuda: Ebola surveillance protocols for arriving travelers Government of Bermuda: Epidemiological surveillance Bermuda’s September surveillance reports documented increased COVID-19 and severe acute respiratory infections, alongside increased reports of gastroenteritis. A later September update continued to report elevated gastroenteritis. Government of Bermuda: Health Information, September surveillance The Bermuda-to-Boston Medical Flight Government of Bermuda: Air ambulance statement The Bermuda government confirmed that an air ambulance departed Bermuda for Boston carrying six people before contact with the aircraft was lost. Government of Bermuda: Air Ambulance statement Associated Press: Crash, passengers and investigation AP reporting identifies Bermudians Marilyn Lavonne Bean and Sergio Wayne Lottimore among the six aboard, describes the reported electrical problems and emergency, and documents the subsequent discovery of debris and suspension of the search. Associated Press: Bermuda passengers identified after medical-flight disappearance Why Boston Wasn’t Actually Strange Mass General Brigham: Medical care for Bermudians Mass General Brigham maintains a dedicated program for Bermudian patients seeking advanced and specialty treatment in Boston. In other words, a medical evacuation from Bermuda to Boston is not inherently unusual. Mass General Brigham: Advanced Specialty Care for Bermudian Patients Boston and Special Pathogens Massachusetts General Hospital: Regional special-pathogens capability Mass General is one of the federally designated regional centers developed to care for patients with Ebola and other high-consequence infectious diseases. Mass General: Regional Ebola and Other Special Pathogens Center Boston University: Current Bundibugyo research Boston University’s Center on Emerging Infectious Diseases has been actively analyzing the current Bundibugyo outbreak, including modeling diagnostic requirements in the DRC. Boston University CEID: Bundibugyo diagnostic research Boston University: Nancy Sullivan and DRC Ebola research BU virologist Nancy Sullivan has longstanding research and scientific-capacity-building relationships in the DRC and published specifically on the 2026 Bundibugyo outbreak. Boston University: Nancy Sullivan Those are the pins. The sources establish the facts. They do not establish the conspiracy theory I constructed from them. That’s the point. The bullshit was never in the pins. It was in the red string.