---
title: "FaceStealer Stole $400M by Faking Faces on Video Calls"
published: 2026-11-15T00:06:55.401Z
updated: 2026-11-15T00:06:55.498Z
author: "Trust Onion Editorial Team"
canonical: https://trustonion.io/blog/facestealer-deepfake-video-call-fraud-doj-seizure
---

# FaceStealer Stole $400M by Faking Faces on Video Calls

**Cite as:** FaceStealer Stole $400M by Faking Faces on Video Calls, Trust Onion, 2026-11-15. https://trustonion.io/blog/facestealer-deepfake-video-call-fraud-doj-seizure

Over 18 months, a North Korean operation called FaceStealer stole more than $400 million by putting AI-generated faces of real, recognizable figures onto video-call screens and convincing developers to install malware.

The contacts arrived through hijacked messaging apps and video-conferencing platforms. On the other side of the call was what appeared to be a known figure from the crypto world, apparently live, apparently speaking. What the targets were actually seeing was an AI-generated deepfake, and what they were installing, at the scheme's direction, was malware.

The US Department of Justice announced the disruption of FaceStealer in November 2026, alongside the seizure of approximately $1.2 billion in digital assets tied to a separate North Korean operation: a $6 billion hack of Binance. The FaceStealer operation alone drained more than $400 million across 18 months. The FBI also seized over 900 webcam videos of victims collected during the scheme.

The case is a clear marker for any organization that conducts business over video. A voice on a phone call is not proof of identity. Neither, now, is a face on a screen.

## How FaceStealer Worked

The scheme did not need to compromise a target's email or guess a password. It needed a convincing enough face and a plausible enough request. Targets, many of them open-source developers, received outreach through platforms they already trusted. Once on a call with what looked like a credible contact, they were guided toward installing software that handed over access.

The deepfakes were built from publicly available video of real people, which is now a routine starting point for this kind of operation. Studies show people often can't tell a cloned voice from a real one, and the same perceptual gap applies to video when the frame rate and lighting are close enough.

For a finance or operations team, the point is not that video calls are unsafe. A face on a screen, like a voice on a phone, is not a sufficient verification step before acting on a high-stakes request.

## What a Verification Layer Looks Like

Before installing software, transferring funds, or changing access credentials on the basis of a video call, a team needs a check that the other side cannot fake. Trust Onion's rotating codewords give teams exactly that: three words, known only to authorized personnel, that change every 60 seconds. A deepfake of a CFO or a known industry contact cannot supply words it has never heard.

The check is simple enough to run in any meeting. Before acting on the request, one team member asks: "What are the words?" If the caller cannot answer, the action stops. No amount of visual or audio realism bridges that gap.

---

The DOJ's seizure of $1.2 billion measures how much state-sponsored operations have extracted through these methods over years. FaceStealer's $400 million came from targets who had no reason, in the moment, to doubt what they were seeing. The FBI's recovery of over 900 webcam videos suggests the operation was logging its victims as it went, a detail that shows how methodical the scheme was and how long it ran before it was stopped.

The organizations that avoided the worst outcomes in cases like this were not the ones with better antivirus software. They were the ones with a step in their process that a convincing face could not complete.

<!-- trust-onion:voice-clone-demo-disclosure -->

## A controlled voice-cloning safety demonstration

[Try our tool to see just how easy voice cloning is](/clone-my-voice). This controlled safety demonstration exists only to show how easy voice cloning is and to make the risk tangible. It is not a real incident or proof of anyone's identity. Your submitted sample is handled transiently to generate the fixed demonstration audio. Trust Onion does not identify you or the person from the sample, and does not keep the submitted recording or generated audio. The temporary voice clone is deleted from our voice provider as soon as the demo finishes; if that deletion fails, it is retried automatically until the provider confirms it.

## Key takeaways

- FaceStealer used AI deepfake video of real public figures to deceive targets into installing malware, stealing over $400 million in 18 months.
- A convincing face on a video call is no longer a reliable identity check for high-stakes business requests.
- Three rotating codewords, known only to authorized team members, give staff a verification step that deepfakes and voice clones cannot pass.

## Frequently asked questions

### What was the FaceStealer deepfake scheme?

FaceStealer was a North Korean operation that used AI-generated deepfake videos of known crypto figures in video calls to trick targets, often open-source developers, into installing malware. It stole more than $400 million over 18 months before the DOJ disrupted it in 2026.

### Can businesses verify someone's identity on a video call?

A face or voice on a call is no longer a reliable identity check. Businesses can add a verification layer by requiring callers to supply a rotating codeword known only to authorized personnel before any high-stakes action is approved.

### How do rotating codewords protect a finance or operations team?

Rotating codewords change every 60 seconds and are known only to people inside the verified group. Before acting on any executive request or installing software, staff ask for the words. A deepfake or impersonator cannot supply them.

### How much did the DOJ seize in the North Korea crypto case?

The DOJ seized approximately $1.2 billion in digital assets linked to North Korean operations, including proceeds connected to a $6 billion hack of Binance, announced in November 2026.

## More from Trust Onion

Give your finance and operations teams a simple two-factor check for high-stakes requests. Trust Onion's three rotating codewords verify the person on the call before any action is taken, free for your whole team.
