CaraComp
CARACOMP DAILY · EP.67

Identity Fraud: Deepfake Family Photos, EU AI Act, Surge · Video Briefing

June 26, 20263:08Watch on YouTube →
Identity Fraud: Deepfake Family Photos, EU AI Act, Surge · Video Briefing
Chapter 1 of 3 · THE VOICE AT THE CLOSING TABLE
0:00 / 3:08

Full Transcript

You're twenty minutes from wiring the biggest payment of your life. Your phone buzzes. It's your loan officer, and the account number just changed. Except your loan officer never called. That's the new face of real estate fraud, and it doesn't target the careless. It targets you on the most stressful day of your year, when you just want to reach the finish line.

THE VOICE AT THE CLOSING TABLE ▶ 0:20

According to Inman Real Estate News, scammers can now clone a working copy of someone's voice from as little as three seconds of audio. Three seconds. And voice phishing attacks jumped more than fourfold in a single year.

A familiar voice calling with urgent, last-minute wire instructions is no longer proof of anything. The only safe move is a callback to a number you found yourself.

That voice felt safe because you knew it. But what happens when the face you trust never filmed the ad at all?

A Korean pop singer named Shin Ji saw an ad featuring a fellow singer she knew personally. Familiar face. Right voice. She trusted it. She shouldn't have, because that performer never made the ad. If a professional who reads faces for a living got fooled, your gut feeling of 'I know them' isn't a safety feature anymore. It's the exact thing being exploited.


THE FACE SHE NEVER FILMED ▶ 1:03

According to Tech Economy, deepfake file production was projected to top eight million files in 2025. That's a sixteen-fold jump in two years. This isn't a basement hobby. It's an industry running around the clock.

Recognizing someone and verifying their endorsement are now two completely different things. And only one of them keeps you safe.

So if knowing the voice and knowing the face both fail, what's actually left to trust?

Picture someone who's studied fakes for years, sitting down to sort real videos from AI-generated ones. They get it right about half the time. Not because they're careless. Because human eyes have hit a wall. The signs we were all taught to look for, the weird blinking, the blurry edges, those clues expired, and most people never got the memo.


YOUR EYES VERSUS A COIN FLIP ▶ 1:47

According to a pre-registered study in Communications of the A.C.M., people scored an average of fifty-one percent spotting fakes. A coin flip is fifty. And people who'd studied deepfakes? They did no better than people who'd never heard the word.

Your eyes are not a verification system anymore. When media asks you to act, confirm the source through a separate channel you already trust.

Three stories, one thread. A voice, a face, a video. Every signal your brain used to trust something has been reverse-engineered. The lesson across all three is the same. Stop asking 'does this look real?' Start asking 'can I confirm who actually sent this?' Verification is about origin now, not appearance. Links to every story and today's podcast deep-dives are in the description. See you next time.

Sources

Stories in This Episode