That Hot Stranger Sliding Into Your DMs? Probably 40,000 Lines of Code.

An account called "Derek Lam" built over 40,000 followers on social media by posting short, silent videos of a gorgeous man dancing. No voiceover. No audio at all — just a face, a body, and a vibe. The face was entirely AI-generated. And tens of thousands of people followed it anyway, many of them genuinely believing a real person was on the other end.

Here's the part that should make you sit up straighter: the scam doesn't start with a request for money. It starts with a flirt. A like. A "hey, you seem interesting." By the time any ask arrives, you've already decided you trust this person — because your brain told you to.

The Fake Profile That Doesn't Need to Be Perfect

According to reporting by Let's Data Science, these synthetic accounts have figured out something clever: skip the audio entirely. No voice. No lip-sync. Just silent vertical video clips — the kind that look completely normal on TikTok or Instagram Reels. That choice is not accidental. Voice matching and lip-sync analysis are exactly the tools researchers use to detect fake videos. By going silent, these accounts sidestep those checks like someone ducking under a laser alarm.

The accounts in question — "Derek Lam," "Leo," and others — weren't small-time operations. Follower counts ranged from 31,000 to 88,000. Some included buried disclosures acknowledging the AI-generated content. Didn't matter. The follows kept coming. The DMs kept going out. And in dating contexts, where these same faces and personas get deployed, the lack of disclosure isn't a bug — it's the whole point.

This is not a story about a niche corner of the internet. It's a story about how attraction has become a weapon. This article is part of a series — start with Your Kids Birthday Photo Is All A Stranger Needs And It Take.

Your Brain Is the Vulnerability

Think about the last time someone attractive messaged you out of nowhere. What happened in your chest? A little flutter, maybe. A quick mental calculation: is this real? Should I respond? And then — almost immediately — a pull toward hoping it was real.

That pull is exactly what romance scammers have always exploited. What's new is that AI has made manufacturing that pull cheap, fast, and nearly undetectable at scale.

Research consistently shows that people cannot reliably tell AI-generated faces from real ones — even when they're specifically told to look for fakes. Detection accuracy falls below what you'd get from a coin flip. Below chance. Meaning that in a direct test, "I think this is AI" and "I think this is real" are basically random guesses. Now take that same person, add attraction, add a flirty opener, add a few days of warm conversation — and the odds of them pausing to verify anything drop even further.

Scammers have a term for the long-game version of this: "pig butchering." (Yes, it's as grim as it sounds.) The idea is that you fatten the pig — build trust, warmth, even emotional intimacy — before you slaughter it financially. What used to require a human scammer investing weeks of real labor can now be automated end-to-end by AI. One operator. Dozens of fake personas running simultaneously. Hundreds of targets at once.

Let that number breathe for a second. Three billion dollars. And that's only what got reported.

Why the Old Detection Tricks Don't Work Anymore

Most people, when they get suspicious, do one of three things: reverse image search the profile photo, ask for a video call, or check how long the account has existed. All three of those methods are now either broken or easily gamed. Previously in this series: Before Your Kid Downloads Another App 28 States Want Your Id.

Reverse image searching only works if the image was stolen from a real person somewhere online. A fully generated face — one that has never existed as a real human — returns nothing. No match. Clean results. Which a scammer can screenshot to show you as "proof" they're real.

Video calls? There are real-time deepfake tools that can overlay a synthetic face onto a live camera feed. The Biometric Update has reported on the rising distrust users feel on dating platforms — and yet distrust alone doesn't tell you what to trust instead.

Account age and follower count? Also gamed. Established fake accounts get sold and repurposed. High follower counts signal credibility even when those followers are also fake.

"Romance scams are now part of the online dating experience — not an edge case but a mainstream risk that users encounter regularly, often without recognizing it." — McAfee 2026 Valentine's Day Research Report

The platform-level safety features haven't kept up. And most users are left trying to make judgment calls with broken tools — like trying to spot a counterfeit bill with no UV light and no training.

What Actually Works — and One Thing You Can Do Tonight

Here's the frustrating truth: the most reliable detection now is behavioral, not visual. A real person behaves differently than a script. Behavioral red flags include: escalation that feels too fast (declarations of connection within days), extreme reluctance to meet in person even after weeks of conversation, and responses that feel emotionally warm but slightly off-topic — like they're answering a slightly different version of what you asked.

According to GRASS, physical presence remains the one verification that AI cannot fully defeat — and notably, Graphika's network analysis of synthetic dating profiles found that 56% of Gen Z users now prioritize arranging in-person meetings before extended digital conversations specifically because of AI trust concerns. They've intuitively arrived at the right answer: meeting in person is currently the only foolproof check. Up next: App Store Age Verification Scotus 28 States.

But here's what you can do before it ever gets that far. If you've ever wondered whether a profile photo or DM is from a real person — that instinct is worth listening to. Run the same face across multiple searches. Look for the same generated features appearing across different accounts with different names. Synthetic faces often share what researchers call "ghost symmetry" — an uncanny smoothness and structural sameness that you can start to spot once you know to look for it. Mismatched earrings, strange hair at the edges, backgrounds that blur in the wrong places.

You don't have to be a tech expert to notice that something feels too perfect. That feeling has always existed for a reason. Trust it more now, not less.

If you've spent time manually staring at suspected fake profiles trying to figure out whether two accounts share the same generated face — drop a comment. You're not being paranoid. You're being right.

The "Derek Lam" account never asked its 40,000 followers for anything. It just danced, silently, accumulating trust. In a dating context, that same face — moved to a private message thread — becomes the setup for whatever comes next. And what comes next is almost never dancing.