Attackers did not crack second factors. They completed them in real time while the employee still believed IT was helping. According to Unit 42 at Palo Alto Networks on November 20, 2026, Com-affiliated actors including Bling Libra and clusters CL-CRI-1264 (Vortex) and CL-CRI-1274 (Cinder) ran sustained workforce vishing from at least August 2026. Callers posing as internal IT support directed staff to organization-tailored adversary-in-the-middle proxy sites that passed passwords and MFA approvals to the attackers live, then used the resulting Microsoft 365 sessions to pull mailboxes, OneDrive, and SharePoint and to send extortion notes over Teams or email.

An adversary-in-the-middle attack is a reverse proxy between the user and the real identity service so the password and the second factor both complete under attacker control. That single login step is what unlocked everything that followed.

How fake IT calls fed tailored AiTM login pages

Operators opened with a voice call that sounded like helpdesk support. The script pushed the worker toward a lookalike portal built for that organization, not a generic lure. Apex domains and short-lived subdomains impersonated SSO portals, IT support pages, and even passkey or MFA registration flows so the page felt routine on a corporate workstation or a personal phone.

Unit 42 stated the servers "act as adversary-in-the-middle (AiTM) proxies that pass user credentials and multi-factor authentication (MFA) approvals to the attackers in real time." The employee typed a password. The employee approved MFA. The proxy forwarded both to the real service and kept the authenticated session for the caller. Centralized SSO and cloud collaboration stacks made one successful ceremony especially valuable: a single workforce identity often reached mail, files, and chat in one hop.

Campaign pressure built through 2026. Unit 42 responded to related data theft and extortion earlier in the year, then saw significant acceleration from at least August. By September, CL-CRI-1264 and CL-CRI-1274 were identified as targeting shifted from financial services toward healthcare. October kept healthcare and financial services at the top of the mix while Australian targeting rose. Public reporting does not name individual victim companies or publish exact victim counts, so the picture stays at industry and country level rather than a clean victim list.

The same Com-affiliated wave also showed a parallel harvest path under CL-CRI-1192 (Helix): a sideloaded browser extension that pulled authentication cookies and MFA tokens, including via QR scans. That path sat beside the core vishing-plus-proxy clusters rather than replacing them. The common theme stayed workforce identity on collaboration platforms, not a consumer loyalty app.

Control in play What operators did Why it failed here
Voice "IT support" trust Coached staff onto fake portals Human trust beat URL scrutiny
Password + MFA at login AiTM relayed both live Second factor was transferable
Durable signed-in sessions Token replay skipped re-prompts Durable session outlived the call
MFA device enrollment Rogue factor added in minutes Stolen session already existed

Rogue MFA devices and durable signed-in token replay

Session capture was only the first durable win. According to Unit 42, attackers behind CL-CRI-1264 enrolled an unauthorized MFA device within minutes of authenticating, including reuse of a device named Pixel 10a across accounts. That move turned a one-time proxied login into a factor the operators could satisfy later without another perfect vishing call.

CL-CRI-1274 took a different persistence cut. Unit 42 reported that those operators used vishing for initial access and avoided re-prompting for Duo MFA by replaying a durable session token that kept the user signed in, captured through the proxy infrastructure. The MFA prompt never returned because the durable session already said the user was signed in. Public detail does not claim identical MFA product behavior for every cluster, so the Duo durable signed-in case stays tied to the cluster Unit 42 described.

Neither persistence move required breaking cryptography on the second factor after the fact. Both required a live, accepted sign-in that legacy phishable login allowed a remote proxy to complete. Once that session existed, enrolling a new factor or replaying a durable signed-in cookie was ordinary abuse of features designed for convenience. Revoking sessions, disabling attacker-registered devices, and hunting Graph abuse become cleanup after the fact. They are not the same as stopping the proxied login.

Graph scripts, mailbox theft, and Teams extortion

With Microsoft 365 sessions in hand, operators stopped needing the employee at the keyboard. Unit 42 described Python tooling built on httpx and requests plus the Microsoft Graph SDK to pull mailbox, OneDrive, and SharePoint content at script speed. Occasional large CRM or recruiting record theft appears in the reporting without published per-victim volumes or payment figures, so those totals stay unknown in the public record.

Extortion followed the data path inward. Messages went out over Microsoft Teams or email, pointing targets at Tor-hosted negotiation sites, so the pressure landed inside the same collaboration tools the business already trusted. That pattern turns a single workforce login into both the theft channel and the shakedown channel.

The authentication failure sits at the start of that chain. A password plus a relayable MFA approval is still a transferable ceremony. A reverse proxy can finish it while the real user believes the page is corporate IT. Device-bound, origin-bound signatures that never send a reusable secret across a fake origin close that remote completion path. After a legitimate session already exists on a machine, stealing it with malware is a harder, different problem. This campaign’s documented entry was the cheaper one: coach the worker, proxy the login, keep the session.

Defenders reading only the extortion note see a data incident. The enabling step was a workforce sign-in that a stranger could complete from a phone call and a tailored proxy. Unit 42’s clusters show how fast that step becomes rogue MFA devices, replayed durable signed-in session cookies, and Graph exfiltration when the second factor can still ride through someone else’s site.

How to prevent this attack is covered in the related article on how device-bound MFA stops AiTM vishing relay on workforce login.

FAQ

How did the Com-affiliated attackers get into Microsoft 365?

The Com-affiliated attackers got into Microsoft 365 by vishing employees while posing as internal IT support and sending them to organization-tailored AiTM proxy sites that relayed passwords and MFA approvals in real time. According to Unit 42, that live relay produced authenticated sessions the operators then used with Graph-oriented scripts against mailboxes, OneDrive, and SharePoint.

Did MFA stop the Bling Libra and CL-CRI-1264 style campaigns?

MFA did not stop the Bling Libra and CL-CRI-1264 style campaigns when the second factor could be completed through an AiTM proxy. Unit 42 documented real-time pass-through of MFA approvals, then separate persistence via unauthorized MFA device enrollment within minutes and, for CL-CRI-1274, replay of a durable session token that kept the user signed in and avoided Duo re-prompts.

What is the Pixel 10a detail in the CL-CRI-1264 reporting?

The Pixel 10a detail is Unit 42’s observation that CL-CRI-1264 enrolled unauthorized MFA devices quickly after proxy authentication and reused a device named Pixel 10a across accounts. That reuse shows operators standardizing persistence factors on top of stolen sessions rather than relying on a single one-off login.

Was this the same as a helpdesk Temporary Access Pass reset?

Public Unit 42 reporting on these Com-affiliated clusters describes vishing that directed workers to AiTM proxy login pages, not a documented helpdesk Temporary Access Pass or genuine-support reset path. The transferable step on the record is the proxied password-plus-MFA ceremony and the sessions that followed.

What data did operators take after the AiTM login?

Operators took Microsoft 365 mailbox, OneDrive, and SharePoint data using captured sessions and Python httpx/requests plus Microsoft Graph SDK automation, then sent internal extortion through Teams or email. Unit 42 did not publish exact victim counts, named companies, per-victim volumes, or ransom payment amounts in the public write-up.