deniz.in

Markets

Weather

Loading weather

· via dev.to (home feed)

DOJ seizes FaceStealer infrastructure behind fake video-call malware and $400M crypto theft

The DOJ disrupted cloud infrastructure behind FaceStealer, a North Korean-linked scheme that used fake video calls and deepfakes to steal over $400 million in cryptocurrency — and it now targets open-source maintainers.

DOJ seizes FaceStealer infrastructure behind fake video-call malware and $400M crypto theft

The US Department of Justice announced on 12 November 2026 that it had disrupted cloud infrastructure behind a North Korean social-engineering operation the FBI calls FaceStealer. According to the DOJ, the scheme used fraudulent video meetings to trick people into installing malware, and it has been tied to the theft of more than $400 million in cryptocurrency over the past 18 months.

What the DOJ announced

Alongside the infrastructure disruption, handled out of the Central District of California, the DOJ said investigators recovered more than 900 webcam videos of FaceStealer victims under a federal warrant. In a separate action covered by the same announcement, over $1.2 billion in cryptocurrency recently stolen from Binance was seized and restrained. The department connected the activity to clusters tracked under names including Sapphire Sleet, UNC1069, Dangerous Password, CryptoCore, APT38, BlueNoroff and Stardust Chollima.

The Romanian outlet Libertatea reported the story on 13 November 2026, adding that the fraudulent meetings were built to imitate Zoom, Microsoft Teams and Google Meet. A technical write-up on dev.to, which walks through the DOJ material, highlights the detail most relevant to developers: the DOJ says the hackers have increasingly gone after people who develop and maintain open-source software, alongside cryptocurrency and web3 workers, with targeting that suggests the aim of enabling supply-chain attacks. An FBI official in Los Angeles is quoted saying the scheme can fool even highly experienced, security-minded technical users.

How the fake call works

The DOJ describes a repeatable sequence. The invitation arrives from a hijacked account belonging to someone the victim knows, so the sender looks familiar. When the victim joins, the attackers record their webcam while playing back pre-recorded footage of earlier victims — possibly people the target recognises — or deepfakes of well-known figures in crypto and web3, so the room appears to contain trusted contacts.

Partway through, a fabricated technical problem appears, typically that the victim cannot be heard, along with a link or prompt to repair it. That repair is the malware payload, installed by the victim's own hands. In several cases the DOJ describes, cryptocurrency theft and account takeovers followed months, sometimes more than a year, after the initial compromise — long enough that nobody connects the loss to the call. The attackers then mine the victim's contact lists and messaging accounts to stage the next round of invitations, and the webcam footage recorded in one call can become the familiar face shown to the next target.

Why maintainers are exposed

A developer laptop frequently holds a live GitHub session, SSH keys, npm or PyPI publish tokens, cloud credentials and CI access. Whoever controls that machine does not need to attack the repository itself — they can publish as the maintainer, pushing a malicious release in front of everyone who installs the package. The DOJ has not published a count of affected maintainers, and the dev.to write-up notes it does not claim any specific person or project has been hit.

The practical defense

The checklist that the dev.to article distils from the reporting requires no new tooling:

  • Before a call: treat invitations as unverified even when they come from friends, and confirm unexpected meeting requests on a separate channel you already trust. Read the full domain of a meeting link rather than its display text, or skip the link entirely and enter the meeting ID in the platform's official app.
  • During a call: never download, paste or run anything offered as a mid-meeting repair. If audio fails, leave, restart the app or reschedule — a legitimate platform does not need you to run commands during a meeting. Treat participants who do not react to what you say or cannot answer an unexpected question as a warning sign, and confirm unusual requests through a channel the attacker does not control.
  • On your machines: keep publish credentials off the device you take calls on, using a separate profile or machine for releases. Use hardware security keys or strong two-factor authentication on GitHub, npm and PyPI, prefer short-lived and narrowly scoped tokens, and apply operating system and browser updates only through each system's own settings.
  • If you already ran something: take the machine offline, then from a clean device change passwords and revoke sessions and tokens for email, code hosting, package registries, cloud consoles and your password manager. Rotate SSH, API, signing and deploy keys, review recent releases and access logs for changes you did not make, warn your contacts, and follow the FBI's published guidance for victims, which the DOJ links from its release.

Why it matters

FaceStealer previews how cheap synthetic and replayed video erodes the trust assumptions built into everyday work: a familiar face on a call is no longer evidence that person is present. The DOJ's supply-chain language signals that open-source maintainers are now deliberate targets, where a single compromised laptop can ripple outward to every downstream user of a package. And the effective countermeasure is not a product but a short set of habits — chief among them refusing to run anything offered as a fix during a call — each of which breaks the attack chain at its weakest link.

  • #security
  • #malware
  • #deepfakes
  • #open-source
  • #cryptocurrency
  • #north-korea

Related posts