Deepfake Attack: Cloned Faces, Voices and Deepfake Video Scams
Quick answer
How can you tell if a video call is a deepfake scam?
Often you cannot tell by looking, because a deepfake can copy a familiar face and voice convincingly. Do not rely on spotting glitches. Hang up and call the person back on a number you already have, or ask something only they would know, before sending money or sharing private details.
Picture a video call from your daughter. She looks scared. She sounds like her. She needs money right now, and she begs you not to hang up. That is a deepfake scam, and Sri Lanka Police just warned the public that it is happening to ordinary people.
A deepfake (a fake video, photo, or voice made by AI to look and sound like a real person) can now fool you on a video call, so a familiar face is no longer proof and you should confirm every urgent request a second way.
Here is the short answer to your 11pm question. Are you safe? Mostly, yes, if you change one habit. The habit is this: stop treating "I recognize them" as the end of the check. Treat it as the start.
What Sri Lanka Police Said About a Deepfake Attack, and Why It Reaches Far Beyond One Country
The warning, reported by Daily Mirror in Sri Lanka, is short and blunt. Criminals are using AI tools to trick people, take their money, and pull out private details. The police did not describe one big case. They described a deepfake method anyone could fall for, and a deepfake attack of this kind needs very little to get started.
Criminals are using sophisticated technology to deceive people, steal money and obtain sensitive information.
Summary of the Sri Lanka Police warning, Daily Mirror (Sri Lanka)
Now the part that should make you sit up. The raw material is already online. Scammers can grab a few photos or a short voice clip from social media and use them to build a deepfake copy of a family member, a friend, or someone you trust. Then they use that deepfake to push you into sending money or reading out your banking details.
Think about what you have posted lately. A birthday video. A voice note in a public group. A clip of your kid at a recital. None of that was posted for a criminal. All of it is usable in a deepfake. This article is part of a series, start with Atlantic City Hotels Own Face Costs Star Hotel Guest 1 250.
How a deepfake call actually works
Nobody needs to be a movie studio anymore. A scammer collects a face and a voice, builds a deepfake copy, and then adds the one ingredient that really does the damage: panic. "Mom, I'm in trouble." "I need this payment before noon." The deepfake face is bait. The deadline is the hook. (Honestly, the deadline is the part I would trust least. Real emergencies survive a thirty-second callback.)
Why Is Deepfake Detection So Hard With Your Own Eyes?
Because your brain is built to trust what it sees and hears. We spent our whole lives treating a familiar face as evidence. A deepfake simply borrows that trust. The research behind this piece cites a test in which only 0.1% of people could correctly pick out deepfake content. That is not a typo, and it is not a knock on those people. It means "look closely for glitches" is weak advice for anyone trying to detect deepfakes.
Even trained eyes struggle. One of the sources put it plainly: deepfake technology has become so real that at times it is tough to figure out, even for a trained eye. So the old deepfake tips (watch the blinking, look for odd skin) are a coin flip at best and a false comfort at worst. Deepfake audio is harder still, because a cloned voice has no skin or blinking to check.
Are deepfake videos multiplying, or is it just in the news?
It is growing, and the numbers are steep. Sumsub tracked about four times as many detected deepfakes in 2024 as in 2023. Pindrop, which watches fraud at company call centers, measured a jump of roughly 1,300%, from about one deepfake attempt a month to about seven a day, according to figures gathered by Keepnet Labs. Seven deepfake attempts a day sounds small until you remember that is one company's phone line.
| What we used to trust | What a deepfake can now fake | What to do instead |
|---|---|---|
| A familiar voice on the phone | A deepfake cloned voice built from a short recording | Hang up and call back on the number you already have |
| A face on a video call | A deepfake video face that moves and talks | Ask something only the real person would know |
| A message from a friend's account | A deepfake or scammer posing as that friend | Confirm with a separate text or an in-person check |
| A request that feels urgent | Deepfake pressure built to stop you from thinking | Slow down and wait before any money moves |
How to detect deepfakes without special software
Look at the right-hand column. None of it needs special software. It needs a pause and a second channel (a different way of reaching the person, like a phone number you already trust). Software-based deepfake detection has a place at banks and platforms, but at the kitchen table a pause works better.
Sri Lanka Deepfake Scams Are a Preview, Not an Exception
It would be easy to read this as a story about one country. Don't. INTERPOL has also flagged the growing use of AI in fraud, where criminals build convincing deepfake videos and voice recordings to pose as real people. The Sri Lanka warning is simply one police force saying out loud what many are seeing. The deepfake scam travels well because the ingredients (a photo, a voice clip, a worried relative) exist everywhere. Previously in this series: Digital Age Check 4 8m Danes Can Skip The Selfie Upload.
Why a deepfake changes the rules for ordinary families
- Recognition is no longer proof. Seeing and hearing someone used to settle the question. Now a deepfake only raises it.
- Your posts are source material. Public photos and short voice clips are enough for a scammer to start a deepfake.
- Speed is the weapon. A deepfake scam works best when you act before you think, so your defense is to buy time.
Can facial recognition tools be fooled by a deepfake too?
Yes. A deepfake or altered face image can push an identity verification system into wrong answers, a risk covered in security research from Mitek Systems on injection attacks (feeding fake images straight into a system) and liveness checks (whether the camera is looking at a live person or a picture). So the problem is not only that people get fooled. Machines can be fooled by a deepfake too, which is why a single glance, human or digital, should never carry the whole decision.
One Thing You Can Do Tonight Against Digital Deception
Pick a family verification step. It can be a code word, a question only your real family would answer, or a rule that nobody sends money after a call until someone calls back on a saved number. Tell your parents. Tell your teenager. Do it before a deepfake call arrives, because panic is a terrible time to invent a system.
If you have ever wondered whether a photo, video, or profile is really who it claims to be, that is the exact question face comparison technology exists to answer. It lines up several images of a face and measures how alike they are, so the judgment does not rest on one gut feeling. That is the same principle you can use at home: never let one look be the whole verdict.
A deepfake turns a familiar face or voice into a tool for pressure, so the safest default is to verify another way before sending money or sharing anything private. As the Sri Lanka deepfake scams warning shows, trusting a familiar face is no longer proof.
Recap: A deepfake can copy faces and voices well enough to fool most people. Check any urgent request through a second channel before you act.
So here is the question to leave on your kitchen table. If your own mother called, crying, asking for money, what exactly would you do in the first ten seconds? If the honest answer is "send it," the people running deepfake scams already know that. Up next: Age Verification 25 States Now Demand Id And Face Scans.
Deepfake: Frequently Asked Questions
What is a deepfake scam?
A deepfake scam uses AI to make a deepfake video, photo, or audio clip that looks and sounds like a real person, often someone you know. Deepfake videos and deepfake audio are both forms of synthetic media. The scammer then uses that deepfake to ask for money, passwords, or banking details. Sri Lanka Police warned that criminals can build these deepfakes from photos and short voice recordings found on social media. The deepfake is only the opening move. The real tool is urgency, which pushes victims to act before they stop to check.
How can I tell if a video call is a deepfake?
Honestly, you often cannot spot a deepfake video call by looking, and an AI-generated video can look flawless on a small phone screen. Research cited in this article found that only 0.1% of people tested could correctly identify AI content, so deepfake detection by eye is unreliable. To detect deepfakes in practice, verify through another channel. Hang up and call the person on a number you already have, or ask a question only the real person could answer. Treat any request for money or private information as unconfirmed until you do.
Why are deepfake scams increasing?
Deepfake technology has become easier to use, and the raw material is everywhere online. Sumsub tracked about four times as many detected deepfakes in 2024 as in 2023. Pindrop measured a 1,300% surge in deepfake fraud attempts at contact centers, from roughly one a month to seven a day. A deepfake attack also grew from 0.1% of global fraud attempts in 2022 to 6.5% in 2025, so criminals clearly see it working.
What did the Sri Lanka deepfake scams warning say?
Sri Lanka Police told the public that criminals are using AI tools to deceive people, steal money, and obtain sensitive information. Scammers can use photos or short voice recordings from social media to build a deepfake of family members, friends, and trusted individuals, then pressure victims into transferring money or disclosing banking details. The message is simple: this digital deception works through a familiar face or voice, so confirm requests another way.
What should a family do to protect against an AI impersonation scam?
Agree on a verification step ahead of time. That could be a shared code word, a personal question, or a rule that money never moves until someone calls back on a saved number. Also think about what you post publicly, since photos and voice clips can feed a deepfake. The goal is to buy time, because these deepfake scams depend on panic. A thirty-second pause and a second channel defeat most of them.
Ready for forensic-grade facial comparison?
Full forensic reports with detailed similarity scoring. Results in seconds.
Run My First SearchMore News
Digital age check: 4.8M Danes can skip the selfie upload
Denmark's biggest payment app can now prove your age, showing only a photo, name and age. Convenient, yes. But is it one more piece of your identity getting tied together?
privacyAge Verification Roblox: Malaysia Weighs ID Checks for Kids
Malaysia is studying age checks for Roblox and other online games. Safety matters, but so does what your family hands over to prove a child's age.
privacyAge verification: 25+ states now demand ID and face scans
Age checks started with adult sites. Now they are creeping toward everyday websites, and the price may be your ID. Here is what to know before you tap "verify."