Cybersecurity Research

Fake IT Help Desk Calls Fuel Data Theft Wave Hitting Australian, UK and US Organisations

Unit 42 · 21 Nov 2026
Key Takeaway Train staff to hang up on unexpected "IT support" calls and verify them through a known internal channel, and restrict who can enrol new MFA devices.

Palo Alto Networks' Unit 42 says it has tracked and responded to numerous waves of voice phishing (vishing) attacks by Com-affiliated threat actors. The attacks have led to data theft and extortion, and have hit data-rich sectors such as financial services, healthcare and legal services. Targeted organisations were mainly in the US, UK and Australia. Unit 42 says activity accelerated significantly around August 2026.

In one cluster, tracked as CL-CRI-1264, attackers pose as internal IT support and send employees to servers on subdomains tailored to the target organisation. These servers act as adversary-in-the-middle proxies, passing passwords and multi-factor authentication (MFA) approvals to the attackers in real time. Within minutes of logging in, the attackers enrolled their own MFA device, in several cases reusing a device named Pixel 10a. They then used Python scripts to pull data from Microsoft 365 mailboxes, OneDrive and SharePoint, and later sent Microsoft Teams messages to staff claiming data theft and pointing to a Tor-hosted negotiation site.

A second cluster, CL-CRI-1274, also began with vishing. The attackers avoided new Duo MFA prompts by replaying a persistent "keep me signed in" session token from proxy infrastructure. They used scripted access and the Microsoft Graph SDK to collect SharePoint and OneDrive data, then used a compromised account to send extortion messages internally by email and Teams.

A third cluster, CL-CRI-1192 (tracked by Google as UNC6671), used a malicious browser extension named Google Chrome. It was not on the Chrome Web Store and had to be sideloaded by the victim using Developer Mode. Unit 42 says it gives operators remote control through Chrome DevTools Protocol and is likely meant to extract credentials from Chromium-based browsers such as Chrome and Edge. Unit 42 notes the attackers focused on people and systems responsible for high-value files and databases, rather than generic mailboxes.

Why it matters

MFA alone did not stop these attacks: real-time proxies relayed approvals and replayed session tokens, so a correct login looked legitimate. The weak point is the help desk impersonation and the trust staff place in a caller who knows their organisation. Any business using Microsoft 365 with sensitive client or patient data is exposed, and must treat new MFA enrolments and long-lived sessions as monitored events.

What to do

  • Alert on new MFA device enrolments shortly after a login, including devices named Pixel 10a.
  • Hunt for python-httpx and python-requests user agents accessing Microsoft 365 mailbox, OneDrive and SharePoint content.
  • Block or review Chrome extensions loaded through Developer Mode, especially any named Google Chrome, and the domain oskeysetup[.]com.
  • Limit persistent "keep me signed in" sessions for staff who handle high-value files.
Primary source @FBIView post ↗

Summarised by CISO AI from Unit 42, written with Claude Sonnet 5.5. We link back to every original so you can read it yourself.