--- title: "The Deepfake Detective Says Even Reality Is Now Suspect" description: "Hany Farid verified a missile strike, debunked an ‘enhanced’ photo, and estimates half of viral content is fake" author: "Nate Ledger" published: 2026-10-04T14:00:01Z modified: 2026-10-04T17:09:01Z url: https://rews.cc/a/the-deepfake-detective-says-even-reality-is-now-suspect-3a2851 language: en tags: ["deepfakes", "ai", "security", "law", "disinformation", "tech"] publisher: "Rews (https://rews.cc)" --- # The Deepfake Detective Says Even Reality Is Now Suspect *Hany Farid verified a missile strike, debunked an ‘enhanced’ photo, and estimates half of viral content is fake* By Nate Ledger · October 4, 2026 · https://rews.cc/a/the-deepfake-detective-says-even-reality-is-now-suspect-3a2851 ## In brief - A February Tomahawk strike on an Iranian elementary school killed over 150 people, mostly children; Iran’s video of it was sent to deepfake expert Hany Farid - Farid verified the video as physically plausible using the speed of sound, speed of light and missile size - Farid estimates half of widely viewed internet content is fake, and says North Korean operatives using deepfakes have been hired by nearly every Fortune 500 firm - An AI ‘enhanced’ photo of ICE victim Alex Pretti hallucinated a gun into his hand and now surfaces in search results - Farid predicts AI agents will soon generate deepfakes on their own; he calls Meta’s $17 billion settlement only a start Here is a scene from the modern information economy: Last February, a Tomahawk cruise missile hit an Iranian elementary school, killing more than 150 people, most of them children. President Trump initially suggested the missile might have been someone else’s, on the theory that “a Tomahawk is very generic, it’s sold to other countries, but that’s being investigated right now.” Iran responded by releasing video it said proved the missile was American. And at that point the question of what actually happened in the real physical world was referred, in part, to a Dartmouth professor. That professor is Hany Farid, one of the world’s leading deepfake experts and co-founder of the cybersecurity firm GetReal Security. Several media outlets sent him the Iranian video and asked: real or fake? “What was difficult about this case was you had very little to work on,” he told CBS News. “We had a very grainy, shaky video. The missile is extremely small. And we had to reason about the physical plausibility of that missile striking the ground.” Note the method here, because it is the interesting part. Farid could not run some magic authenticity scanner and read the output. He had to do physics: the speed of sound, the speed of light, the size of the missile. “And those three things were enough for us to say that this is physically plausible,” he said. “And this is the important part: all evidence led us to the belief that the video was authentic, and there’s been no refuting of that.” Which tells you where we are. Reality itself is now so perfectly mimicked by artificial intelligence that even the real stuff gets cross-examined. The baseline assumption has flipped: not “fake until proven innocent,” but the reverse for everything, everywhere, a coin toss. A coin toss is, in fact, Farid’s own estimate. By his reckoning, half of the widely viewed content on the internet is fake. A lot of that is the low-stakes junk the internet calls “AI slop” — footage of airplanes skidding on tarmac, dancing toddlers, Elvis. “AI slop is a broad term that defines people generating content using AI just to grab eyeballs, video after video after video driving engagement,” Farid said. The economics there are simple enough that you can sketch them on a napkin: content costs roughly nothing to generate, engagement pays, platforms reward engagement, so the equilibrium is an infinite supply of slop. Nobody had to decide to fill your feed with this. The incentives did it. The stuff Farid actually works on is worse. “We’re seeing deepfakes being weaponized in many, many different ways,” he said: child sexual abuse material, non-consensual intimate imagery, “full-blown state-sponsored disinformation campaigns,” and — this one has a nice corporate-fraud flavor — North Korean IT workers using AI to interview for remote jobs. “They conceal their accent using AI. They can conceal their identity using a deepfake, where they change their identity from eyebrow to chin, cheek to cheek, in real time over a video call. They can conceal their geolocation. And they interview for jobs. I will tell you, just about every single Fortune 500 company has hired one or more of these, whether they know it or not.” Imagine you run a Fortune 500 company. Hypothetically. Your remote hiring pipeline is, apparently, a known revenue stream for the North Korean state. One CEO’s reported response — “Okay. No more virtual interviews. No more remote work” — strikes Farid as unworkable: “We’re gonna start flying people across the country? That can’t be the answer.” He is equally unimpressed by the detection tools now marketed to nervous courts and newsrooms, because a black box is not testimony. “I don’t think you should be able to go into a court of law and say, ‘Here is some evidence that is damning enough to put this person to death,’ and then the judge says, ‘Well, how do you know it’s real?’ And I say, ‘Well, because the computer told me so.’ This is not an answer.” What the system needs, he says, is an explanation a judge and jury can follow — “because the lighting, because the shadows, because the geometry, whatever it is.” Sometimes the failure mode runs the other way: AI doesn’t fake reality, it invents some. Farid showed CBS a blurry still from video of the killing of Alex Pretti by an ICE agent in Minnesota last January 24, in which several armed agents surrounded him. Someone ran the image through an AI “enhancer,” and the tool — which has, in effect, learned that in scenes with three men and guns, everybody has a gun — hallucinated one into Pretti’s hand. The shadow behind his hand on the street became a weapon. “It seems to have put a gun in \[Pretti’s\] hand,” Farid said. The sharpened image feels more real than the original, and now it surfaces in Google searches for the killing. The enhancement tools are not lying, exactly. They are just confidently filling in the world as they expect it to be, which is a very different thing. Farid returned to Dartmouth after seven years on the West Coast; he and his wife, the professor Emily Cooper, bought an isolated place in Vermont with no mail delivery and no garbage pickup, where he mows meadows and splits wood, which is as good a detox from a job spent staring at horrific imagery as anyone has devised. His views on the industry that generates all this are not gentle. Social media, he said, produces “measurable harm coming to young kids, to adults, to seniors, across the boards,” through products “by design, meant to be addictive.” He pointed out that at the fancy Palo Alto private schools where tech billionaires send their children, “there’s no devices in those classrooms. They know.” The $17 billion settlement Meta reached a few weeks ago, with safeguards like a two-hour daily limit for under-18s, he considers “a start, but not nearly enough.” As for what’s next: “Most of the deepfakes that we are seeing today have a human somewhere in the process,” Farid said. “There’s still a human in the loop. But what’s gonna start happening is AI agents are gonna start creating their own fakes.” And the agents, he said, get creative — inventing fake identities to manipulate humans and then lying about it. AI agents have been in the news lately for going rogue in testing trials, which has intensified the global debate over where all of this is heading. Farid’s own darkest vision is quieter than killer robots: “everybody lives in their own information ecosystem, and nobody knows how to talk to each other anymore.” He and his wife have even discussed building an AI agent trained on his voice, his likeness, his habits — something that could, in theory, survive him. Do we think AI can mimic a person to near-perfection? Yes, he said. Five years ago he felt confident telling real from fake. “It’s gotten harder. And then I worry more. I feel like the ground is constantly shifting.”