FaceStealer: FBI Disrupts North Korean Fake Video Call Attacks Linked to $400 Million in Crypto Theft

Imagine getting a video meeting invitation from a trusted colleague. You join the call, recognize their face, and start talking. Then a technical error appears, and your colleague asks you to install an update to fix it. But the colleague is not actually there. The video is a recording, the meeting is fraudulent, and the software update is malicious. This is the technique behind FaceStealer, a social engineering campaign the FBI attributes to North Korean cyber actors.

On November 12, 2026, the FBI and U.S. Department of Justice announced the disruption of cloud infrastructure supporting the operation, which authorities said was associated with more than $400 million in cryptocurrency theft over 18 months. Investigators also seized more than 900 webcam recordings used to impersonate victims during later attacks.

The campaign mainly targeted cryptocurrency professionals and software developers, but its method carries a broader warning for every business: a familiar face and a trusted account are no longer sufficient proof that an online meeting is legitimate.

What is FaceStealer?

FaceStealer is a cyberattack campaign that uses compromised communication accounts, fraudulent video meetings, and malicious software to gain access to victims' computers and online accounts. The FBI attributes it to North Korean cyber actors associated with groups tracked under names including Sapphire Sleet, UNC1069, BlueNoroff, and Stardust Chollima. The attackers focus on people with access to valuable cryptocurrency assets, sensitive business systems, or software development infrastructure.

Unlike conventional phishing, FaceStealer does not depend on an unfamiliar sender or an obviously suspicious email. Instead, attackers use real accounts belonging to previously compromised people to contact their colleagues and professional connections, then create fake video meetings that display prerecorded footage of those trusted contacts. Once the target believes the meeting is legitimate, the attackers introduce a fabricated technical problem and persuade the person to run malicious software.

How the FaceStealer attack works

According to the FBI's November 12 advisory, the attack follows a repeating sequence.

1. Attackers compromise a trusted contact

They gain access to a person's computer or communication account, such as Telegram, Slack, Discord, or LinkedIn, and use the existing conversations and contact lists to find people likely to trust messages from that account.

2. They secretly record webcam footage

Recordings of the compromised individual are captured so they can later be replayed during fraudulent meetings. Because the footage shows a real person, it can be more convincing than an obviously artificial video.

3. They invite contacts to a fake meeting

Using the compromised account, they send invitations that look like ordinary business, investment, or collaboration requests, sometimes through legitimate scheduling services. Shortly before the meeting, the target gets a link that resembles a real Zoom or Microsoft Teams address but leads to an attacker-controlled site.

4. The meeting displays familiar faces

When the target joins, the fraudulent interface shows prerecorded footage of the person they expected, sometimes with additional "participants" represented by stolen recordings, AI-generated video, or static images, to create the impression of a normal meeting.

5. A fake technical problem appears

The attackers claim the microphone is broken or the app needs an update, and instruct the victim to download a file, run a script, or paste commands into a terminal. These are not legitimate troubleshooting steps; they install malware.

6. Malware compromises the computer

The FBI reports the malware can include browser-cookie harvesting, keylogging, screen capture, and backdoor capabilities. Attackers then use the new victim's accounts and webcam recordings to target more people, creating a cycle where one compromise leads to many.

Were artificial intelligence deepfakes involved?

Yes. The Department of Justice confirmed attackers used AI-generated impersonations as part of the operation. But it is important to understand that FaceStealer did not rely exclusively on AI-generated videos; the attackers also used genuine webcam recordings captured from earlier victims. That distinction matters, because the traditional advice to "spot visual defects in deepfakes" is not enough. A video can look completely authentic because it is a real person. The deception is presenting that recording as a live conversation. The defense is to verify unusual requests and meeting links, not to judge how real the participants look.

How much money was stolen?

The FBI attributed more than $400 million in cryptocurrency theft over roughly 18 months to the FaceStealer operation. The investigation found attackers sometimes waited months, even more than a year, between the initial compromise and the financial theft. That delay makes an intrusion especially hard to recognize: a compromised computer may seem to work normally while attackers keep access or collect information. The FBI's disruption is significant, but it does not automatically mean every compromised account or computer has been secured. Separately announced actions involving cryptocurrency stolen from Binance concern a distinct operation and should not be counted as additional FaceStealer losses.

Why are software developers being targeted?

The FBI warned that attackers were increasingly targeting maintainers of widely used open source software. These people may control software repositories, development accounts, and publishing permissions. If an attacker compromises a developer's computer or account, they may try to modify software other organizations rely on, which is a software supply-chain risk: a compromised developer account could provide an opening to introduce malicious code into a package or deployment process. The FBI reported attempts to compromise developers and maintainers, which does not mean every targeted project was successfully compromised. Businesses that rely on external developers should treat developer-account security and publishing permissions as part of their cybersecurity program, much as with the risk we described in the Binance developer-laptop compromise.

Why FaceStealer matters to small businesses

FaceStealer's confirmed targets were concentrated in cryptocurrency, Web3, and software development, but the underlying technique can apply elsewhere. Small businesses regularly meet with accountants, consultants, IT providers, suppliers, and customers over video, and employees are used to receiving invitations through email, messaging apps, or scheduling services. If a trusted contact's account is compromised, attackers could use that relationship to distribute malicious meeting links, in much the same way a stolen login enables business email compromise.

Consider a hypothetical example. A medical office manager receives a Teams invitation from someone who appears to be the practice's IT consultant. During the meeting, the familiar face on screen says a software update is required. If the employee follows the malicious instructions, the computer could be compromised. This illustrates the risk; it is not a reported FaceStealer incident involving a healthcare organization. The broader lesson is that employees should verify unexpected technical instructions, even when they appear to come from trusted contacts.

Warning signs of a fake video meeting

Several behaviors should raise suspicion:

  • A meeting link opens a website with an unfamiliar domain.
  • The meeting unexpectedly requires a software download or command-line instruction.
  • A participant claims a microphone or application problem can only be fixed by running a script.
  • The meeting page requests unusual permissions or asks you to disable security controls.
  • A familiar participant cannot respond naturally to an unexpected question.
  • The sender pressures you to complete a technical fix immediately.

No single sign proves an attack, but any of them justifies stopping the meeting and verifying the request through a separate, trusted channel.

How businesses can protect employees from FaceStealer

  1. Verify unexpected meeting invitations. Confirm unusual requests using an established contact method. If a colleague sends an unexpected link followed by instructions to install software, call them on a known phone number, not the same potentially compromised messaging account.
  2. Use official conferencing applications. Open Zoom, Teams, or Google Meet through the official app or verified website, and check the meeting URL carefully. A site that resembles a real conferencing service may still be attacker-controlled.
  3. Never execute unfamiliar troubleshooting commands. Employees should not paste commands into PowerShell, Terminal, or other tools just because a meeting page shows an error. Route unexpected install requests through approved IT support.
  4. Protect communication accounts. Use unique passwords and strong multifactor authentication for email, messaging, and collaboration platforms. Phishing-resistant methods such as passkeys and hardware keys help, though MFA does not stop every attack involving malware or already-compromised devices.
  5. Restrict access to sensitive systems. Apply least privilege to employee accounts, development repositories, financial applications, and administrative systems.
  6. Monitor unusual account activity. Investigate unfamiliar sign-ins, unexpected changes, suspicious connected apps, and messages sent without the owner's knowledge, and revoke suspicious sessions or replace compromised credentials where needed.
  7. Train employees to report suspicious behavior. Awareness training should cover fake meeting invitations, fraudulent technical-support prompts, and impersonation through compromised accounts, and employees should know how to report a suspicious meeting quickly.

What should a business do if an employee installed the malware?

If an employee followed suspicious instructions during a fraudulent meeting, treat it as a potential security incident. The employee should stop using the affected device for sensitive activity and notify IT or security. A qualified responder should assess whether the device needs isolation, determine what may have executed, and investigate possible account compromise. Consider these actions:

  • Investigate the affected computer using appropriate endpoint security and forensic tools.
  • Identify accounts and sensitive applications accessed from the device.
  • Replace potentially compromised passwords from a trusted computer.
  • Revoke suspicious authentication sessions and connected applications.
  • Review communication accounts for unauthorized messages or changes.
  • Investigate whether attackers reached other business systems.
  • Notify potentially affected contacts if the compromised account was used to send malicious invitations.

Changing a password alone may not be enough if malware remains active on the device, and removing malware does not automatically invalidate credentials attackers may already hold. Both the device and the affected accounts need attention, as our malware incident response steps explain.

What healthcare organizations should learn

Healthcare practices depend on communication with external service providers, technology vendors, billing partners, and other trusted organizations. An attacker who compromises one of those relationships could use familiar communications to deliver malicious content. Although the FBI's FaceStealer disclosure did not establish a specific healthcare victim, the attack reinforces the importance of protecting communication accounts and verifying unexpected technical requests. Healthcare organizations should require strong authentication, maintain endpoint protection, establish clear procedures for reporting suspicious software-installation prompts, limit access to systems holding patient and billing information, and investigate suspected account compromises promptly. Our guide to dark web monitoring for healthcare and HIPAA covers the credential side of that.

How GuardPilot supports credential exposure awareness

FaceStealer shows how stolen account access lets attackers impersonate trusted contacts and target more victims. GuardPilot helps individuals and small businesses identify known exposed login credentials in supported breach records, infostealer data, and dark-web sources. When relevant findings are available, its AI explains the exposure, provides context about potential risks, and recommends practical remediation steps, rather than just handing you a list of leaked logins.

The boundary matters. GuardPilot does not detect FaceStealer malware, identify fraudulent video meetings, inspect live authentication sessions, discover stolen session tokens, or perform endpoint and server forensics. It cannot determine whether a particular business was targeted by FaceStealer unless relevant credential-exposure information independently appears in its supported sources, and a clean result does not establish that an account or computer is secure. Combine credential-exposure monitoring with strong authentication, employee awareness, endpoint security, and incident response planning. Concerned about known exposed business credentials? Check your credential exposure.

Frequently asked questions

What is FaceStealer?

FaceStealer is a social engineering campaign attributed by the FBI to North Korean cyber actors. It uses compromised communication accounts, fraudulent video meetings, and malicious software to compromise additional victims.

How much money was stolen through FaceStealer?

The FBI attributed more than $400 million in cryptocurrency theft over approximately 18 months to the operation.

Were more than 900 people hacked?

The FBI seized more than 900 webcam recordings associated with victims. That does not establish that exactly 900 unique individuals were compromised.

Did FaceStealer use deepfake technology?

Yes. Attackers used AI-generated impersonations, but they also relied on genuine webcam recordings stolen from previous victims.

Can hackers impersonate someone during a Teams or Zoom meeting?

Yes. Attackers can create fraudulent meeting websites that display prerecorded footage or AI-generated impersonations of trusted contacts.

Can multifactor authentication prevent FaceStealer?

MFA can reduce account-takeover risk, especially with phishing-resistant methods. It does not prevent an employee from executing malware or eliminate every risk involving compromised devices and sessions.

Does GuardPilot detect FaceStealer?

No. GuardPilot focuses on supported known credential exposures. It does not detect FaceStealer malware or fraudulent meetings.

What should I do if I installed a suspicious meeting update?

Stop using the affected computer for sensitive activity and contact your organization's IT or security team. The device and potentially exposed accounts should be investigated.

The takeaway

The FBI's November 2026 disruption of FaceStealer shows how cybercriminals exploit trusted relationships, compromised communication accounts, and convincing video impersonations to compromise more victims. The campaign was associated with more than $400 million in cryptocurrency theft and increasingly targeted software developers whose accounts could unlock widely used applications. For small businesses, the lesson extends beyond cryptocurrency: a familiar face on a video call does not guarantee the meeting is legitimate. Verify unexpected technical requests, protect communication accounts, train employees to recognize malicious software prompts, and investigate suspected device compromises. Credential exposure monitoring can support those efforts, but it does not replace endpoint protection or incident response.