Deepfake Scams Cost Singapore S$242.9M as Deepfakes Surge

Here's a number that should stop you mid-scroll: S$242.9 million. That's what people in Singapore lost to government impersonation scams in 2025, more than double what they lost the year before. Some of those calls came with a fake video of a senior official's face, moving and talking, built by AI using deepfakes. The government's answer isn't more warnings. It's giving official calls a recognizable prefix, a short code at the start of the number, so your phone can flag "this one's really us" before you even pick up.

A deepfake scam wave that cost Singapore residents S$242.9 million in a single year has pushed the government to pilot a common prefix on official calls, a small signal meant to help people spot real government outreach before AI-generated voices and faces trick them into acting fast.

What is a deepfake scam, and how do deepfakes change the sound and look of government calls

A deepfake scam is when someone uses AI to fake a person's face, voice, or both, well enough to fool you into thinking you're talking to someone real. Not a grainy photo. A moving, talking, convincing fake made from deepfakes. Scammers in Singapore have been using this deepfake attack pattern to impersonate government officials in video calls, showing a fabricated face of someone who supposedly works for a real agency, telling victims they're under investigation or owe money or need to move funds "for safekeeping." Government impersonation scam cases hit 3,363 in 2025, and the average victim lost S$72,229. That's not pocket change. That's a car, a year of rent, someone's retirement cushion. Deepfake phishing calls often layer a fabricated voice on top of that fake video, so the scam deepfake feels doubly convincing, hitting both your eyes and your ears with fabricated data at once.

So Singapore's answer, according to Fintech News Singapore, is a pilot program where government agencies will use a shared, recognizable prefix on outgoing calls later in 2026. Think of it like a verified checkmark, but for your ringing phone instead of a social media profile. If it works in testing, it rolls out wider across agencies. The idea is simple: give residents one more real signal to check, something a scammer can't easily fake, before they decide whether to trust the voice on the other end. This kind of identity check matters because scams increasingly rely on stolen or synthesized data rather than a scammer's own voice or face.

How deepfake scam calls and voice cloning actually trick people

Here's the part that should genuinely unsettle you: scammers only need about 3 seconds of someone's voice to build a clone with 85% accuracy. Three seconds. That's shorter than most voicemail greetings. Combine that with an AI-generated face on a video call, and you've got a fake official who looks like he's really there, sounding exactly like he should, telling you to wire money right now before your account gets frozen. Deepfake fraud attempts have jumped 2,137% over three years, according to research cited by StationX. That's not a slow creep. That's an explosion. Video deepfakes, voice cloning tools, and other deepfake schemes are getting cheaper and faster to produce, which is exactly why these scams keep multiplying and why every employee handling sensitive data or wire transfers at a company is now a potential target too, not just individual residents.

---

Why the phone call is no longer proof of who's speaking amid rising deepfake scams and deepfakes

For most of your life, hearing someone's voice on the phone was basically proof enough. Sounds like your mom, it's your mom. Sounds like your bank's fraud department, it probably is. That assumption is dead now, and Singapore's prefix pilot is basically the government admitting it out loud. Voice alone and video alone are no longer authoritative, full stop. If a machine can clone a voice from three seconds of audio and slap a fake face on a video call, then the thing you used to trust most (the actual sound of a human speaking to you) is now the easiest thing to fake. This is the core threat behind nearly every deepfake scam video making the rounds right now, whether it targets a resident, a business, or a government worker handling sensitive data. These deepfakes are the reason old habits no longer hold up.

This is why Singapore's own messaging leans on what officials describe as a multi-layered approach, meaning no single fix, not even a fancy prefix, is going to stop scam calls on its own. It takes banks flagging weird transfers, platforms cracking down on fake profiles, agencies rolling out signals like this prefix, and regular people pausing before they act. Take away any one layer and the other three have to work harder. Security teams inside banks and telecoms now treat every unusual video request as a potential fraud attempt until proven otherwise, and that shift in security posture is spreading well beyond Singapore.

The pilot only works if carriers and law enforcement can actually prevent spoofing of the prefix number, which remains technically difficult and inconsistently enforced.

Expert analysis on Singapore's common prefix pilot, Fintech News Singapore

And that skepticism is fair. Spoofing a phone number costs fractions of a cent and works against plenty of carrier systems. A scammer could, in theory, try to fake the exact prefix Singapore rolls out. Whether the pilot actually holds depends on how tightly it's paired with anti-spoofing enforcement at the network level, not just a nice-looking code on your screen. A prefix without enforcement is a costume, not a lock. Previously in this series: Taylor Swift Deepfake Sexual Images Pulled In 17 Hours Podca.

Deepfake scam prefix pilot versus your old caller ID habits

People have relied on caller ID for decades as a rough gut check. A government prefix is meant to be a step up from that, a signal that's supposedly harder to counterfeit because it ties back to verified agency phone systems rather than whatever number a scammer typed into a spoofing app. But "harder to counterfeit" isn't "impossible to counterfeit," and that gap is exactly where your own judgment still has to do the heavy lifting against a growing wave of scams and fraud.

Old caller ID trust	Common prefix pilot	Status
Relies on whatever number displays on screen	Ties to a shared, agency-verified code	Piloting 2026
Easily spoofed for pennies per call	Harder to fake, but not spoof-proof without network enforcement	Enforcement pending
Voice and video treated as proof on their own	Voice alone and video alone are no longer authoritative, so the prefix is only a first layer against deepfakes	Ongoing security gap
No callback protocol expected	Still requires hanging up and verifying independently amid rising identity threat	Recommended now

---

Beyond phone calls, the same deepfake tools power a growing category of video scams aimed at businesses. A scam video showing a fake executive requesting an urgent wire transfer has already cost companies real money in other countries, and Singapore's prefix pilot is partly a response to that same underlying identity threat. Once you understand that a deepfake attack can hit a phone call, a video meeting, or even a job interview, the prefix starts to look less like a gimmick and more like one small piece of a much bigger security puzzle that touches personal accounts, company payrolls, and government identity checks alike.

The everyday stakes of scams built on deepfake scam calls and deepfakes, and what you can actually check

So what does this mean for you, tonight, if your phone rings and something feels off? It means an official-looking prefix or caller ID is a helpful clue, not a verdict. If you've ever wondered whether the "government official" on your screen is really who they claim to be, that's the exact question this kind of verification signal exists to help answer, but it can't answer it alone. Here's one thing you can actually do, tonight: if any call demands money, passwords, one-time codes, or urgent action, hang up and call the agency back using a number you find yourself, from an official website or a bill you already have. Not a number the caller gives you. Not a callback button on a text they sent. One you looked up yourself, cold.

Here's where it gets interesting, and honestly a little grim: enterprises that adopted call authentication tools saw spoofing attacks drop by 28%, but international spoofing still slips past local protections more than half the time, according to data cited by Stingrai. So even a well-built system leaks. That's not a reason to ignore the prefix pilot. It's a reason to treat it the way you'd treat a smoke detector, useful, worth having, but not a substitute for actually checking the stove. Fraud teams that combine call authentication with employee training on deepfake attacks tend to catch far more attempts than technology alone ever could, especially as deepfakes keep improving.

What comes after the deepfake scam pilot in Singapore

Expect this to spread. Once one government proves a common prefix reduces confusion, even a little, banks and telecoms elsewhere will copy the idea, the same way multi-factor authentication went from "nice extra step" to "just how logging in works" after the 2021 wave of ransomware attacks. This is table stakes, version one, not the finished fix. The finished fix, if there ever is one, will need real cooperation between phone carriers on anti-spoofing standards, something that has stalled in a lot of countries because it requires everyone to move at once.

---

Picture the moment right before you answer. Your screen shows a prefix that's supposed to mean "trust me." The voice on the other end sounds exactly like the official it claims to be, urgent, official, maybe even a little kind about your "situation." Everything about that call was built to feel like proof. None of it is. The most honest thing Singapore's pilot admits is that the burden hasn't shifted to your phone at all, it's still sitting right there in your hand, waiting for you to hang up and dial the number yourself. Up next: Biometric Based Authentication A Face Is Just 512 Numbers Po.

deepfake scam and deepfakes: Frequently Asked Questions