Deepfake Scams: One Fake Video Call Cost Her $287,000

A California woman lost $287,000 to a man she never actually met — because the video calls, the voice, the face on her screen, were all fake. That's what deepfake scams look like now: not a blurry photo and bad spelling, but a person who blinks, smiles, and says your name back to you in real time. An Australian man's story, aired on A Current Affair, is the same playbook with a different face — one AI-generated "girlfriend" or "boyfriend," months of trust, and money gone before anyone figured out the person didn't exist.

Here's the part that should worry you, whether you're single, married, or just have a lonely parent who's active on Facebook: the old advice doesn't work anymore. Reverse-searching a photo used to catch scammers, because stolen model pictures show up everywhere online. It doesn't catch a face that was never a real photo to begin with — it was built by an AI, pixel by pixel, and it doesn't exist anywhere else to be found.

The New Math Behind Deepfake Scams

For years, romance scam warnings focused on volume. Watch for the guy messaging fifty women at once. Watch for stock photos. Watch for grammar mistakes that scream "overseas call center." That advice made sense when scammers were running numbers games — cast a wide net, hope a few bite.

Deepfake technology broke that math. Now one fake person, built well, can hold a video call, answer questions in a voice that sounds warm and specific, and keep a relationship going for months. According to research from the Alan Turing Institute, scammers now use large language models — AI programs trained to write like a real person — to generate messages tailored to each victim's interests and emotional weak spots. That's not a bot sending copy-paste lines. That's a system reading your replies and adjusting, the way a good con artist always has, except now it never gets tired and never runs out of patience. This article is part of a series — start with How To Spot A Deepfake.

How Deepfake Romance Scams Actually Work Now

A scammer builds one convincing AI identity — face, voice, backstory — then uses it in real-time video calls and voice messages to build trust over weeks or months, before ever asking for money.

That number comes from Euronews, and it's not a typo — losses jumped from around €304 million in 2024 to $1.1 billion in a single year. Some of that is crypto fraud and fake investment pitches. But a growing slice is romance and companionship scams, where the "product" being sold isn't a stock tip — it's affection. In Australia alone, people reported more than 3,200 romance scams in 2024, losing close to $24 million, according to the Australian Banking Association. And that's just what got reported — shame keeps a lot of victims quiet.

Why Deepfake Scam Losses Keep Climbing

Losses aren't just growing because there are more scammers. They're growing because each scam works better than it used to. A con that once took a scammer months of typing now runs mostly on autopilot, with AI handling the emotional labor of "getting to know you." Researchers at UNSW studying AI-enabled romance fraud found that these systems are built specifically to exploit trust and loneliness — not by accident, but by design.

Modern romance-scam operations use AI systems to generate emotionally tailored messages around the clock, responding within seconds and adjusting to each victim's personality and vulnerabilities — the kind of round-the-clock attention a real relationship rarely offers this early. — reporting summarized from KnowBe4

That's the trap, honestly. A real person gets busy. They take a day to text back. They have bad moods. An AI-run "partner" is always available, always fascinated by you, always exactly what you need to hear — which feels incredible right up until it feels like a warning sign nobody taught you to notice, because "too attentive" has never been on anyone's scam checklist. Previously in this series: Cctv Facial Recognition.

Can You Actually Tell If Someone Is a Deepfake?

Not reliably by eye anymore. Live deepfake video can mimic blinking, lighting, and voice well enough to fool most people on a phone screen. The only real defense is refusing to send money, private images, or personal details until you've verified the person through a channel outside the relationship itself.

If you've ever had that flicker of doubt about someone you met online — the photo that seemed a little too polished, the video call that felt slightly off, the person who was available at 3 a.m. every single time — that instinct is worth trusting. That's the exact question this kind of technology exists to answer: is the face on your screen actually a real, consistent person, or something built to look like one? One thing worth doing before you get emotionally or financially invested: ask for an unscripted, unplanned video moment — like turning their head fast, or holding up a specific object right then — because that kind of on-the-spot request is still one of the harder things for a live deepfake to fake convincingly.

Researchers at UNSW point out that romance scam victims often experience psychological effects on par with grief or trauma — not surprising, since the relationship felt entirely real to them while it lasted. The money is the headline number. The harder part to recover from is realizing months of intimacy were manufactured, on purpose, by someone who studied exactly what you needed to hear. Up next: How To Spot A Deepfake 1 School Photo Is All It Takes.

The A Current Affair case will get watched, shared, maybe forgotten by next week's news cycle. But the pattern underneath it isn't going anywhere: deepfake scams don't need an army of fake profiles anymore. They need exactly one good one, aimed at exactly the right lonely evening. The scariest part isn't the technology — it's how ordinary the setup looks right before it works.

Frequently Asked Questions