When Deepfakes Hijack Reality
The digital underworld can bend your eyes, your ears, and maybe your future
I remember the exact moment I stopped trusting video calls. Not completely — I still take them, obviously — but there’s this new reflex now, this tiny hesitation before I believe what I’m seeing. It happened after I read about the Hong Kong case.
Last year, a finance worker at a multinational firm transferred $25 million to fraudsters. The kicker? He’d attended a video conference with his company’s CFO and several colleagues first. Everyone looked right, sounded right, moved right. Except every single person on that call was a deepfake.
I keep thinking about that worker. Does he replay that meeting in his head, searching for the tell he missed?
This isn’t science fiction anymore. It’s just… Tuesday.
We’ve crossed a threshold most people haven’t noticed yet. The technology to fabricate convincing video and audio of anyone saying anything — it’s out there now, available to anyone with a laptop and moderate determination. While we’ve been busy arguing about whether AI will take our jobs, this quieter catastrophe has been unfolding underneath: we can no longer trust what we see and hear. And honestly? I’m not sure we’ve processed what that actually means yet.
When Footage Becomes Fiction
The Hong Kong heist is just one flavor. In the States, elderly parents keep getting panicked calls from their “children” — voices perfect down to the specific way their kid says “Mom,” cloned from a TikTok video posted three months ago. Scammers need three seconds of audio. Three seconds.
A CEO of a UK energy company — smart guy, been in business for decades — transferred €220,000 because he thought he was talking to his boss. The voice had everything: the slight German accent, that characteristic pause before ending sentences, even the impatience when explaining things twice. The money vanished into accounts that don’t technically exist.
But here’s where it gets worse. (Does it always get worse with this stuff? Anyway…)
During Slovakia’s 2023 elections, a fake audio recording dropped two days before voting. A liberal candidate, allegedly caught discussing vote-rigging with a journalist. It spread everywhere — WhatsApp, Facebook, those forwarded-by-your-uncle chains. By the time fact-checkers debunked it, millions had already heard it. The election results were razor-thin.
In India, deepfake videos of actresses flood social media constantly now — their faces grafted onto explicit content they never filmed. Rashmika Mandanna woke up one morning to find herself “starring” in a viral video she had zero part in creating. For ordinary people — teachers, teenagers, that guy from accounting — the stakes are even higher. One video. That’s all it takes to end a career, destroy a reputation, obliterate a life.
Should we be building a world where your face can betray you without your consent? I genuinely don’t know.
Apps That Turn Anyone into a Counterfeit Star
Walk into any college dorm and mention deepfakes. Students will pull up FaceSwap, Reface, Zao — apps that let you insert your face into movie scenes with a few taps. Mostly harmless. Upload a selfie, become Iron Man for thirty seconds, send it to your group chat. Everyone laughs.
The same technology powers something else entirely, though.
Telegram channels and Discord servers trade “premium” apps — that’s what they call them, premium, like they’re offering extra features instead of removing safety guardrails. These tools create revenge porn, fabricate compromising situations, manufacture evidence of events that never happened.
A high school in New Jersey discovered male students had used one such app to create nude images of female classmates. Just… took their Instagram photos and undressed them. The girls didn’t pose for these images. They didn’t consent. They simply existed online. That was enough.
Similar cases keep emerging — Spain, South Korea, Australia. The geography doesn’t seem to matter much.
Dating apps have become hunting grounds. Scammers create profiles using AI-generated faces of people who literally don’t exist but look perfectly, unnervingly real. They build relationships, earn trust over weeks or months, then request money for emergencies that never happened. Romance fraud cost Americans $1.3 billion in 2022. Deepfakes have industrialized the process — scaled it up like some grim startup.
Meanwhile, Tom Hanks didn’t promote that dental plan. Taylor Swift didn’t recommend that weight-loss supplement. But millions of fans see the videos, assume they’re real, click through to scam websites. The celebrities spend their time issuing denials instead of doing actual work. What a way to spend your career.
The Viral Clips That Fooled Millions
Remember Pope Francis in that stylish white puffer jacket? March 2023, the image spread everywhere — Twitter, Instagram, actual news sites. People debated his fashion choices. Some praised him. Others criticized the Church for extravagance.
Nobody noticed: the Pope never wore that jacket. A construction worker in Chicago created it on Midjourney, just experimenting with the tool. He didn’t even mean for it to go viral.
That same month, images of Trump being arrested by New York police circulated — dramatic shots, him resisting officers, being pinned to the ground. They looked like breaking news captured by someone’s phone. Complete fabrications, created before any actual arrest occurred.
A video of Zelenskyy allegedly telling Ukrainian soldiers to surrender went viral in March 2022, right after Russia’s invasion. The deepfake was crude by today’s standards — the head didn’t quite match the body, the voice had that uncanny quality — but in the chaos of war, who’s checking frame-by-frame? Thousands shared it.
Actually, no — I should be more precise here. The deepfake was crude by late-2024 standards. By March 2022 standards, it was pretty convincing. That’s the thing — the bar keeps moving.
More recently, a deepfake of Baltimore’s mayor using racist language spread through local social media. City officials scrambled. Relationships between the mayor’s office and various communities fractured in real-time. The video was fake. The consequences? Those were real enough.
How Law Enforcement Tries to Keep Up
Police departments weren’t built for this, and it shows.
Get JM Bonthous’s stories in your inbox
Join Medium for free to get updates from this writer.
A detective in California described the challenge to a reporter: “Someone reports a deepfake video destroying their reputation. We know it’s fake. They know it’s fake. But how do we prove it in court? How do we find who made it? The creator could be anywhere in the world, using a VPN, working through anonymous accounts that lead nowhere.”
Some jurisdictions are fighting back. Texas passed a law in 2023 making it illegal to create deepfakes intended to influence elections. California banned deepfake pornography created without consent. Virginia criminalized the distribution. But enforcement? Nearly impossible. By the time authorities identify a source, dozens of new fakes have already launched.
The FBI created a Deepfake Task Force in 2023. Agents admit, off the record, that they’re outgunned. For every deepfake they identify and remove, dozens more appear. It’s like bailing out a boat with a teaspoon while someone’s drilling new holes in the hull.
Courts are struggling too — maybe even more than law enforcement. How do you authenticate video evidence anymore? Defense attorneys now routinely claim body camera footage, security recordings, cell phone videos could all be deepfakes. Even legitimate evidence gets challenged. A judge in Florida recently spent three days having experts verify that a surveillance video was authentic before allowing it as evidence. Three days for something that used to take three minutes.
Some police departments are adopting digital signature technology — embedding cryptographic markers in their body camera footage to prove authenticity. But this only helps with new recordings. Older evidence, civilian phones, footage from before these systems existed? All vulnerable to doubt.
What Happens When Victims Fight Back
Sarah — not her real name, she asked me to change it — discovered her face on pornographic videos spreading through her industry’s professional networks. Marketing exec in Seattle. Colleagues sent concerned messages, that awful mix of genuine worry and barely-concealed curiosity. Her employer’s HR department called a meeting. The videos weren’t her. The damage was immediate anyway.
She hired a reputation management firm. Fifteen thousand dollars. They issued takedown notices to hundreds of websites. Most ignored them, especially sites hosted in countries with weak enforcement. Her lawyer sent cease-and-desist letters to the creators. Or tried to — the accounts were anonymous, the trail cold.
Sarah spent six months fighting shadows. Some platforms removed the content. It kept reappearing under new URLs, new accounts, like playing whack-a-mole with her own reputation. She couldn’t apply for jobs without wondering if potential employers had seen the videos. Two years later, the content still surfaces occasionally.
“You can’t un-ring that bell,” she told me over coffee last month. Her hands were wrapped around the cup like she was cold, even though the café was warm. “The internet has a perfect memory. Mine, apparently, does too.”
Public figures have slightly more resources. When Tom Hanks discovered those deepfake dental plan videos, his legal team mobilized — formal warnings to platforms, public statements, threatened lawsuits. The videos came down from major platforms. But they still circulate in darker corners of the web, places his lawyers can’t quite reach.
The Race to Build Authenticity Tools
Tech companies are scrambling. Microsoft released Video Authenticator in 2020, analyzing subtle pixel-level details to detect manipulations. Intel created FakeCatcher — clever name — which detects deepfakes by analyzing blood flow in faces. Real humans show subtle color changes in their skin that current deepfakes can’t replicate.
The problem? These tools work brilliantly on yesterday’s deepfakes, struggle with today’s, and will probably be useless against tomorrow’s. It’s an arms race, and I’m not entirely sure which side is winning.
Meta implemented a policy requiring political ads to disclose synthetic media use. YouTube announced similar requirements. Enforcement depends on creators self-reporting, though. Right. That’ll work.
Content authentication systems might be more promising — emphasis on might. Adobe’s Content Credentials technology embeds cryptographic information into images and videos at creation, building a chain of custody. Canon announced cameras that digitally sign photos as they’re captured. News organizations are piloting systems that would let readers verify whether footage came from authenticated sources.
The Coalition for Content Provenance and Authenticity — Adobe, Microsoft, Sony, the BBC, others — is working to establish industry standards. Their vision: a future where legitimate media carries verifiable credentials, letting audiences distinguish authentic content from synthetic fabrications.
But here’s the catch. These solutions only work if everyone adopts them. A professional news outlet might authenticate its footage, sure. But what about the citizen journalist filming a protest on their phone? What about the whistleblower sharing documents? Verification systems could become gatekeeping tools, accidentally determining whose voices get heard and whose get dismissed as potentially fake.
The Hong Kong finance worker who transferred $25 million — I wonder if he still takes video calls. I wonder if he looks for the tiny glitches, the almost-imperceptible lag, searching for proof that the person on screen is real this time.
We’re building a future where “seeing is believing” has become genuinely dangerous advice. Where reality itself requires authentication. Where your face and voice might betray you without your permission, your knowledge, your consent.
Are we ready for this? Does it matter if we’re not?
The deepfakes are already here.
Three questions for you:
- If we can’t trust video evidence anymore, how do we prove anything actually happened? What fills that gap in courts, in journalism, in daily life?
- At what point does the burden of proof flip — when do we start assuming everything is fake until proven real, instead of the other way around?
- Who decides which verification systems become standard, and what happens to voices that can’t or won’t authenticate through those systems?
Browse past entries or subscribe to the full series here → https://articles.jmbonthous.com/
- Jean Marie Bonthous (publishing as JM Bonthous) is the author of The AI Culture Shock and six other books on the human side of artificial intelligence. His writing also explores truth, authority, and belief, along with the intersections of AI, nonfiction filmmaking, adult learning, and marketing. He examines how technology transforms human experience and how societies decide what to believe.
- He also blogs on Medium about Digital Art: https://medium.com/@jmbonthous
- See his latest books:
www.jmbonthous.com - AI IN REAL-LIFE™ — Connect with us on MEDIUM and follow @jmbonthous1 to stay in the loop with the latest stories about the human side of AI. Let’s help shape the human side of AI together!
