The Voice on the Phone Sounds Exactly Like Them. It May Not Be Them.

Deepfakes now account for 6.5% of fraud attempts, up from 0.1% in 2022, with at least $3.7 billion in documented losses. How voice and video cloning scams target families and finance teams, and the simple checks that stop them.

Cloned-voice scams rely on panic: a familiar voice, an emergency, and a demand for money right now. Photo: Ehimetalor Unuabona, via Wikimedia Commons (CC0)

By the UISC BD Editorial Desk · United Information Service Center · Published 13 September 2026 · 6-minute read

A panicked call from a relative who needs money now. A video meeting where the finance director asks for an urgent transfer. Both used to be reasonably safe to believe. Neither is any more.

How Fast It Has Grown

Deepfakes make up 6.5 percent of fraud attempts, up from 0.1 percent in 2022.
An estimated 8 million deepfakes are expected in 2026, compared with about 500,000 in 2023.
At least $3.7 billion in losses has been documented, 89 percent of it from 2025 to the first half of 2026.
Fewer than 5 percent of voice-clone victims report the crime, so the real figure is higher.

In 2025 the FBI also began tracking AI-related crime as its own category, logging more than 22,000 complaints and about $893 million in losses — part of the wider picture in our report on the FBI's 2025 figures.

Why Voice Cloning Is So Effective

Current tools can produce a voice match of around 85 percent from just three seconds of audio. A short social media video, a voicemail greeting or a recorded talk can be enough.

The attack works because it targets trust and urgency at the same time. A familiar voice in distress short-circuits the instinct to check.

The Business Version

For companies, the losses per incident are far larger. The average voice deepfake incident costs around $600,000, and 23 percent of cases exceed $1 million.

The best-known case came in Hong Kong in January 2024, when an employee of the engineering firm Arup joined a video call with what appeared to be the company's chief financial officer and other colleagues. Every other participant was a deepfake. The employee made 15 transfers totalling about $25 million.

The Family Version

Consumer deepfake scams are usually simpler: a cloned voice claiming an accident, an arrest or a stolen phone, and a request for money by transfer, gift card or crypto. The same technology now appears in romance and investment fraud, where fake live video calls help convince victims — see how pig butchering scams use AI.

Checks That Work

For families:

Agree a code word that only close family knows, and ask for it in any urgent request for money.
Hang up and call back on the number you already have for that person.
Ask a question only the real person could answer — not something visible on social media.
Treat any demand for gift cards or crypto as a near-certain scam.

For businesses:

Require two people to approve large or unusual payments.
Verify requests through a separate channel — never through the call, email or chat that made the request.
Make it safe to question senior staff. Deepfake fraud relies on employees not wanting to challenge the boss.
Train finance teams with real examples like the Arup case.

Can Detection Tools Catch It?

Banks and security firms are deploying AI to spot synthetic voices and faces, and some of those defences are covered in our guide to how banks use AI against fraud. But detection is an arms race. The most reliable protection is a process that does not depend on recognising a voice or face at all.

Sources

"Deepfake statistics and trends," Keepnet Labs — keepnetlabs.com
"Deepfake fraud losses 2026," BR Side — brside.com
"Deepfake statistics 2025," DeepStrike — deepstrike.io
"2025 IC3 annual report," FBI Internet Crime Complaint Center — ic3.gov

voice cloning scam (1)

Deepfake Fraud: Three Seconds of Audio Can Clone a Voice