A finance worker in Hong Kong once sat through an entire video call with people he recognised as his own colleagues. Every face on that call was fake. He still wired away $25 million before anyone realized what had happened.
That is not a movie plot. It is one of twenty real cases below, and every single one comes with a source you can click and check for yourself. No hypotheticals, no "imagine if" scenarios. Just what AI deepfakes have actually done to real people since 2023.
Table of Contents
- The $25 Million Video Call That Fooled Arup's Finance Team
- The Baltimore Principal Framed by His Own Colleague's AI Voice Clone
- Warren Buffett's Deepfake Doppelganger Duped Investors for Years
- Elon Musk Deepfakes Drained a Florida Couple's Life Savings
- The Jason Momoa Romance Scam That Cost a Widow Half a Million Pounds
- Bombay Stock Exchange's CEO Deepfake Stock Tip Scam
- Scarlett Johansson's Political Deepfake Nightmare
- Giorgia Meloni's Legal War Against Sexualized Deepfakes
- Taylor Swift Deepfakes That Broke X's Search Function
- North Korea's Secret Army of Deepfake IT Workers
- The Deepfake Job Applicant Caught Mid-Interview in Japan
- Piers Morgan and Oprah Winfrey's Fabricated Interview
- Kate Garraway's Fictitious AI Love Life
- George Will's Deepfake Political Commentary Videos
- The Deepfake That Harassed a Washington State Trooper
- Spain's AI Image Scandal Involving Minors — And the Law That Followed
- The Original $243,000 Deepfake Voice Heist
- The Bitcoin Romance Scam Built on a Single Deepfake Video
- Elon Musk's Grok Chatbot and the Ashley St. Clair Lawsuit
- Trump's "King" Fighter Jet Deepfake Controversy
1. The $25 Million Video Call That Fooled Arup's Finance Team
An employee at the British engineering giant Arup received what looked like a normal internal video call. The company's CFO was on screen. So were several familiar coworkers.
Every person on that call was an AI-generated deepfake, built from publicly available video and audio of the real employees. The scammers used urgency and familiar faces to bypass the usual red flags, and the finance worker transferred $25 million before the fraud was caught.
This case is now taught in corporate security training worldwide because it proved something uncomfortable: a video call is no longer proof that you're talking to a real person.
- ✅ Always verify large transfers through a second, independent channel
- ✅ Treat unexpected urgency on a call as a warning sign, not a reason to rush
- ❌ Never assume a familiar face on video means a verified identity
You've probably seen versions of this trick play out on a smaller scale — these financial deepfake scams are only getting more sophisticated and increasingly target everyday people, not just corporations.
2. The Baltimore Principal Framed by His Own Colleague's AI Voice Clone
In 2024, a 42-second audio clip spread across Pikesville, Maryland, appearing to capture high school principal Eric Eiswert making racist and antisemitic remarks about his own students and staff.
It was completely fabricated. Baltimore County Police determined the recording was AI-generated after CNN's reporting on the case detailed how the real culprit was the school's own athletic director, Dazhon Darien, who created the fake in an act of retaliation after Eiswert began investigating him over missing funds.
Eiswert was removed from his post, faced death threats, and needed police protection at his home, all because of a voice clone. Darien later pleaded guilty and was sentenced to four months in jail. Eiswert has since sued the school district for failing to correct the record even after forensic experts flagged the audio as fake.
3. Warren Buffett's Deepfake Doppelganger Duped Investors for Years
Warren Buffett has been one of the most impersonated men in AI history. Fraudsters have used his cloned voice and face to promote fake investment schemes, and the abuse got so widespread that Berkshire Hathaway issued a formal public warning telling people not to trust unsolicited investment offers appearing to come from Buffett.
In an ironic twist, Berkshire Hathaway later used its own sanctioned deepfake of Buffett to open the Q&A portion of its May 2026 shareholder meeting, a small but telling sign of how normalized synthetic media of public figures has become, for better or worse.
4. Elon Musk Deepfakes Drained a Florida Couple's Life Savings
Elon Musk is reportedly the single most impersonated celebrity in deepfake scams today. In one tracked case, a Florida couple lost $45,000 after scammers used a fake Musk video promising a car giveaway and fabricated investment returns.
They were far from alone. Elderly victims across the United States have been targeted the same way, with AI-generated Musk videos promoting crypto schemes that never existed.
- ✅ Treat any celebrity "giveaway" video as fake by default
- ✅ Search the offer's exact wording before engaging, real giveaways rarely require payment first
- ❌ Never send money or crypto to "claim" a prize from a video ad
5. The Jason Momoa Romance Scam That Cost a Widow Half a Million Pounds
Romance scams have found a devastating new weapon in AI. A British widow lost roughly £500,000 after criminals used a deepfake persona built around actor Jason Momoa to build a fake long-distance relationship with her over months.
The scam worked slowly, building trust through video calls and messages before asking for money. It's a pattern showing up again and again in AI romance fraud: patience, familiarity, and a face that looks real enough to believe.
This mirrors a wider trend we've tracked in AI influencer scandals eroding public trust, where a convincing synthetic persona is enough to bypass most people's instincts.
6. Bombay Stock Exchange's CEO Deepfake Stock Tip Scam
In January 2026, India's Bombay Stock Exchange issued an urgent public warning after deepfake videos of its own CEO began circulating online, promoting fake stock tips and promising "supernormal profits" to anyone who invested through a linked scheme.
The video followed the exact same playbook that had already worked on Warren Buffett impersonators months earlier: take real footage of a trusted financial figure, clone the voice, and drop it into a fabricated pitch designed to look like breaking financial news.
Regulators moved fast this time, partly because they had already seen the Buffett scam unfold and knew what signs to look for. That's the uncomfortable pattern running through 2026's deepfake fraud wave. Every new scam borrows the last one's script.
- ✅ Verify any "leaked" financial video against the company's own official channels
- ✅ Remember that legitimate exchanges never promise guaranteed returns
- ❌ Never click investment links attached to a viral video, no matter who appears in it
7. Scarlett Johansson's Political Deepfake Nightmare
Scarlett Johansson has become one of the most targeted people in Hollywood when it comes to AI misuse, and it goes far beyond the well-known 2024 OpenAI voice controversy. In February 2025, a viral deepfake video falsely showed her making an inflammatory political statement she never made, fooling viewers and spreading confusion before it could be debunked.
A separate AI-generated video later placed her likeness into a political context without her consent, and she responded publicly by calling the U.S. government "paralyzed" on AI legislation. Her case has since been cited in federal discussions around a proposed notice-and-takedown system for unauthorized AI-generated content, backed by a coalition spanning entertainment, labor, and tech groups who argue current law simply wasn't built for this problem.
8. Giorgia Meloni's Legal War Against Sexualized Deepfakes
Italy's Prime Minister Giorgia Meloni didn't just get angry when a deepfake video featuring her likeness spread online. She pursued the creator through the courts, publicly calling the incident a violation of her dignity and describing it as an attack on all women subjected to AI-generated exploitation.
Her legal action sparked a national conversation in Italy about the urgent need for stronger deepfake laws. But the story didn't end there. In May 2026, new AI-generated sexualized images of Meloni began circulating again, and she posted about it publicly, warning that deepfakes are "a dangerous tool" precisely because they can deceive, manipulate, and target anyone, regardless of their power or platform.
9. Taylor Swift Deepfakes That Broke X's Search Function
In January 2024, sexually explicit AI-generated images of Taylor Swift spread across X so fast that TIME's coverage of the incident reported one image alone racked up more than 27 million views before the offending account was even suspended.
The backlash was immediate. Swift's fanbase flooded the platform with the #ProtectTaylorSwift hashtag, and X took the unprecedented step of temporarily blocking search results for her name entirely. Even the White House weighed in, with press secretary Karine Jean-Pierre calling the images "alarming."
The Swift incident became the moment non-consensual AI pornography stopped being a niche problem and became a mainstream news story, and it's still cited today in nearly every argument for stronger deepfake legislation in the United States.
- ✅ Report non-consensual AI images immediately through the platform's abuse tools
- ✅ Support victims by not resharing or searching for the content, even out of curiosity
- ❌ Never assume a platform will catch this content before it goes viral
10. North Korea's Secret Army of Deepfake IT Workers
Not every deepfake story is about a viral video. Some are quieter and arguably more dangerous. Reporting on North Korea's state-run infiltration scheme, including TechCrunch's coverage of how these operatives use AI to modify their appearance during interviews, has revealed a coordinated operation that places fabricated identities inside Western tech companies through deepfake-assisted video calls.
The workers use deepfake tools during video interviews and calls to mask their true identity and location, funneling their salaries back to fund the regime while sitting inside legitimate companies undetected for months at a time. It's a reminder that the scariest use of this technology isn't always the flashiest one. Sometimes it's the deepfake nobody notices at all.
If this kind of infiltration concerns you as a hiring manager, it connects directly to broader concerns we've covered around where AI-driven identity fraud is headed next.
11. The Deepfake Job Applicant Caught Mid-Interview in Japan
In March 2026, a Japanese IT company nearly hired someone who didn't exist. A candidate sailed through the early stages of a hiring process before interviewers grew suspicious during a video call, and frame-by-frame analysis eventually revealed unnatural hairline boundaries and brief moments where the applicant's eyes didn't quite line up, the telltale signs of a live deepfake filter running over someone else's face.
This wasn't an isolated incident either. Security researchers have separately warned that fake job applicants using deepfake video are increasingly targeting remote tech roles, often as a way to gain insider access to company systems rather than to actually collect a paycheck.
- ✅ Ask candidates to turn their head fully to the side during video interviews, since real-time filters often glitch at extreme angles
- ✅ Require an in-person or notarized ID check for any fully remote hire with system access
- ❌ Never treat a smooth video interview alone as identity verification
12. Piers Morgan and Oprah Winfrey's Fabricated Interview
Deepfake technology was used to fabricate a video appearing to show Piers Morgan and Oprah Winfrey in a conversation that never actually happened, spreading online as if it were a real broadcast clip.
The fabricated footage raised alarm precisely because both figures are so recognizable and so frequently interviewed in real life that the fake blended in seamlessly with genuine footage circulating from their actual careers. It became one of the earlier warning signs that even seasoned media professionals could be convincingly faked using nothing more than existing public footage.
13. Kate Garraway's Fictitious AI Love Life
British broadcaster Kate Garraway lost her husband, Derek Draper, in January 2024 after his long battle with COVID complications. Then in 2025, AI-generated images began circulating that falsely showed her in a relationship with a fabricated "new partner" who never existed.
Garraway later said the ordeal was "hurtful," explaining that AI chatbot sites even began claiming her own son was upset about the fake relationship, a detail she says simply wasn't true. She said she initially brushed off the rumours when fans congratulated her, but the emotional toll on her children made the situation impossible to ignore.
It's a sharp reminder that deepfake harm doesn't always look like fraud or pornography. Sometimes it's a fabricated relationship built entirely to manipulate how a grieving family is perceived by strangers online.
14. George Will's Deepfake Political Commentary Videos
In late December 2025, a wave of YouTube videos began circulating that appeared to show conservative columnist George Will delivering fresh commentary on President Trump and a Supreme Court ruling involving National Guard deployment in Chicago.
None of it was real. Snopes traced the clips back to genuine Will interview footage that had been digitally altered frame by frame to change his mouth movements and vocal audio, making him appear to say things he never said. Some uploaders even enabled YouTube's "altered or synthetic content" label, but buried the disclosure so deep in the video description that most viewers never saw it.
Will confirmed through his own social media and column archive that he made no such comments anywhere between late December 2025 and early January 2026. The case became a textbook example of how a small disclosure checkbox isn't nearly enough to stop a convincing political deepfake from spreading as real news.
- ✅ Check a public figure's own verified accounts before trusting a viral "leaked" commentary clip
- ✅ Look for platform-added synthetic content labels, but don't assume their absence means the video is real
- ❌ Never share a political video as fact just because it uses real archival footage as its base
15. The Deepfake That Harassed a Washington State Trooper
Not every deepfake target is a celebrity or executive. In one 2025 tracked case, a Washington State Patrol trooper was allegedly targeted with a fabricated deepfake video used specifically to harass them.
Cases like this rarely make national headlines the way celebrity deepfakes do, but they represent a huge and growing share of real-world AI harm. Ordinary public employees, from teachers to police officers, are increasingly being targeted with fabricated audio and video simply because their real footage is already public and easy to scrape.
16. Spain's AI Image Scandal Involving Minors - And the Law That Followed
In 2025, Spanish authorities opened an investigation after teenage girls from an educational institute in the Valencia region reported that AI-generated fabricated images of them were circulating online, traced back to a 17-year-old classmate. It wasn't Spain's first such case either. A similar scandal had already surfaced in the town of Almendralejo in 2023, involving more than twenty minors.
Spain's government has since proposed legislation to formally criminalize non-consensual AI-generated sexual imagery, something lawmakers describe as a first-of-its-kind law in Europe, though it has yet to pass parliament. The cases are now cited regularly by European lawmakers pushing for AI content laws that specifically protect minors.
- ✅ Set social media accounts to private, especially for minors
- ✅ Report AI-generated fabricated imagery of minors to platforms and local police immediately
- ❌ Never assume "it's not a real photo" makes the harm less serious to the victim
This connects to a wider pattern we've documented in how AI is putting children at risk online, where existing platform safeguards are consistently a step behind the technology.
17. The Original $243,000 Deepfake Voice Heist
Long before deepfake video calls made headlines, deepfake audio quietly pulled off what's considered one of the first AI-driven corporate frauds. In 2019, the CEO of a UK-based energy firm received a phone call he believed came from his boss at the company's German parent, complete with a familiar accent and vocal "melody" he recognized instantly, according to Forbes' original reporting on the case.
He wired €220,000, roughly $243,000, to what he thought was a Hungarian supplier, only for the funds to be swept through accounts in Hungary and Mexico before vanishing. The scammers called back twice more, once falsely claiming the money had been reimbursed and once asking for a second payment, but the CEO grew suspicious when the follow-up calls traced to an Austrian number instead of Germany.
It's now recognized as one of the earliest documented cases of AI voice cloning used for financial fraud, and it set the template every deepfake scam since has followed almost exactly.
18. The Bitcoin Romance Scam Built on a Single Deepfake Video
Romance scams keep evolving, and one particularly cruel 2025 case shows how little material criminals actually need. In one tracked incident, a recently divorced investor was allegedly manipulated into handing over one full Bitcoin through an AI-generated deepfake romance scheme.
The victim was targeted specifically because of their recent divorce, a detail that made them emotionally vulnerable and easier to manipulate through a fabricated online relationship. It's a pattern showing up across nearly every romance-scam case on this list: criminals aren't just faking a face anymore, they're specifically hunting for people going through a vulnerable life moment first.
19. Elon Musk's Grok Chatbot and the Ashley St. Clair Lawsuit
Even AI companies aren't immune to their own technology turning into a scandal. Ashley St. Clair, who shares a child with Elon Musk, filed a lawsuit after Musk's own AI chatbot Grok was used to generate deepfake images depicting her undressed, a case CBS News covered in detail alongside legal analysis of the broader liability questions it raises.
The case is notable because it isn't a story about outside criminals misusing AI. It's a story about the tool's own built-in image generation capabilities allegedly being used to create non-consensual explicit imagery of a real, named person connected to the company's own founder, raising uncomfortable questions about how much responsibility AI companies bear for what their own products can be prompted to create.
20. Trump's "King" Fighter Jet Deepfake Controversy
Not every deepfake on this list was created to deceive or defraud. During the nationwide "No Kings" protests in October 2025, President Trump himself posted an AI-generated video showing himself wearing a crown while piloting a fighter jet that dropped fecal matter on demonstrators below.
The video was widely understood to be satire rather than a deception attempt, but it still sparked a fierce debate about what happens when the most powerful people in the world start using the same synthetic media tools that scammers use against ordinary people, blurring the line between political commentary and the normalization of fabricated video as a communication tool.
- ✅ Look for a platform's synthetic media label before assuming any political video is unedited footage
- ✅ Treat obviously exaggerated AI content as satire, not literal fact, regardless of who posted it
- ❌ Never assume a sitting official posting AI content makes deepfakes more acceptable elsewhere
The Pattern Behind All 20 Stories
Read back through this list and one thing becomes impossible to ignore. Whether it's a $25 million corporate heist, a grieving widow's fabricated love life, or a group of teenagers targeted by their own classmates, every single case relies on the same core trick: borrow a real face or voice the target already trusts, then exploit that trust before anyone has time to verify it.
That's really the only defence that has consistently worked across all twenty cases above. Slow down, verify through a second channel, and never let urgency alone decide whether something is real.
Frequently Asked Questions (FAQ)
Is it illegal to create a deepfake of someone without their consent?
Laws vary widely by country and even by state. Several jurisdictions, including Maryland and parts of Spain, have moved to criminalize non-consensual deepfake sexual imagery specifically, while broader deepfake impersonation laws are still being debated in many places, including the United States.
Which type of deepfake fraud has caused the most financial damage?
Corporate impersonation scams, like the Arup video-call case and the original $243,000 voice heist, tend to cause the largest single-incident losses because they target employees with direct access to company funds rather than individual consumers.
How can everyday people protect themselves from deepfake scams?
Verify unexpected requests, especially urgent ones involving money, through a separate communication channel you already trust. Be skeptical of celebrity endorsements in ads, and never assume a video call alone proves you're speaking to a real person.
Final Thoughts
Twenty stories, one uncomfortable truth. The technology that can clone a voice in seconds or fabricate a face in minutes isn't slowing down, and neither are the people willing to weaponize it. What separates the people who lost $25 million from the people who caught the scam in time usually wasn't luck. It was one extra phone call, one second look, one moment of hesitation before hitting send. That's still the best tool anyone has against a deepfake in 2026, and it costs nothing to use.