Attackers Use Passkey Phishing to Hijack Microsoft Cloud Accounts and Exfiltrate Knowledge

Microsoft has disclosed particulars of two campaigns during which risk actors are abusing third-party electronic mail supply infrastructure to blast monetary fraud rip-off messages and utilizing passkey-themed social engineering to breach cloud environments.

The primary marketing campaign, per the tech large, concerned sending over one million rip-off emails between August 3 and 5, 2026, by masquerading as chief govt officers (CEOs) of assorted goal corporations, aiming to influence accounts payable departments at these corporations to provoke Automated Clearing Home (ACH) transfers for a supposed ServiceNow annual subscription.

Proof signifies that the operators behind the marketing campaign have leveraged generative synthetic intelligence (AI) to facilitate the creation of electronic mail templates and draft emails tailor-made to their recipients. The exercise primarily singled out enterprise customers within the U.S., spanning IT companies, client items, actual property, and discrete manufacturing sectors.

“The marketing campaign follows steps earlier than and throughout the execution of the marketing campaign: risk actors register impersonation domains, ship executive-themed cost requests by way of trusted infrastructure, embed fabricated invoices and supporting conversations, and try and persuade finance personnel to provoke ACH transfers,” the Microsoft Safety Analysis workforce stated.

“In contrast to conventional bill scams that depend on a single social engineering lure, this marketing campaign layered govt impersonation, vendor branding, fabricated invoices, and supporting electronic mail conversations right into a unified narrative meant to scale back recipient skepticism.”

The spoofed electronic mail messages contained a purported “approval” of the pretend bill to trick recipients into making funds to attacker-controlled accounts. To lend a veneer of legitimacy to the deception, the risk actor included a solid electronic mail thread together with the fabricated bill.

The second marketing campaign documented by Redmond revolves round cloud-based intrusions focusing on a number of accounts during which suspicious sign-ins are adopted by the risk actors including their very own authentication strategies, in addition to high-volume Microsoft Graph exercise, SharePoint and OneDrive downloads, and mailbox assortment by way of REST APIs.

The exercise, which has been detected since Could 2026, is in keeping with “automated assortment from compromised cloud identities utilizing proxy-associated infrastructure,” Microsoft stated.

The assault generally begins with identity-focused social engineering. The risk actors name or message a person’s private telephone quantity, whereas claiming to be from the group’s IT assist desk and urging them to right away replace their passkey, multi-factor authentication (MFA), or single sign-on (SSO) configuration to keep away from entry disruptions.

Unsuspecting staff are redirected to counterfeit web sites that mimic the professional Microsoft sign-in expertise through SMS messages despatched to their private units. The top purpose right here is to make use of the pretext to information them by way of adversary-in-the-middle (AitM) or device-code authentication flows and take management of their Microsoft accounts both by capturing the credentials or unknowingly granting entry on the actor’s behalf.

“The actor seems to take a position closely in pre-attack analysis, seemingly gathering details about staff and organizational construction from public sources similar to social networking {and professional} profiling platforms,” Microsoft stated. “In a smaller variety of circumstances, actors reap the benefits of already compromised accounts to increase their attain” by sending related passkey-themed messages through Microsoft Groups.

What’s extra, the risk actor has been noticed registering domains constructed round themes similar to passkeys, SSO enrollment, account activation, and identification verification, on the identical time together with the goal group’s identify as a subdomain within the sample: “.[.]com” –

• passkeyhelpdesk[.]com
• secure-passkey[.]com
• setupmypasskey[.]com
• add-passkey[.]com
• integratedsso[.]com
• oktasession[.]com
• syncmykey[.]com
• portalsetuphub[.]com

It is value noting that this modus operandi overlaps with a loose-knit cybercrime collective tracked by the cybersecurity neighborhood underneath the monikers Cordial Spider, O-UNC-045, PREY-0058, and UNC6671. The e-crime adversary has been described as a coordinated group of risk actors that operates a number of public extortion manufacturers whereas sharing overlaps within the underlying phishing infrastructure and focusing on footprint.

“UNC6671 makes use of credential harvesting panels hosted on generic root domains masquerading as being associated to passkeys, appending victim-specific subdomains to facilitate focused voice phishing campaigns,” famous final month.

Though the precise nature of those connections is unclear, it is suspected that they’ve been pushed by splintered associates retaining entry to shared preliminary entry playbooks or counting on the identical commoditized phishing panels, voice-phishing callers, and shared infrastructure.

Microsoft, for its half, has attributed the preliminary entry exercise noticed on this marketing campaign to a spread of risk actors, together with Storm-3121 and Storm-3032. Whereas Storm-3121 carries out preliminary entry exercise resulting in ShinyHunters and Falcon (aka CL-CRI-1182) extortion, Storm-3032 is its designation for UNC6671, which refers to a set of actors that broke off from the BlackFile (aka CL-CRI-1116) group and now function underneath the Helix extortion model. 

In not less than one case investigated by Microsoft, the risk actors are stated to have carried out an anomalous sign-in to Microsoft Workplace Residence from an unmanaged machine to increase their entry to different purposes like SharePoint On-line and OneDrive by way of the Graph API and enumerate delicate recordsdata and inside companies.

One other incident concerned the usage of a passkey lure to launch a tool code phishing assault and acquire management of a sufferer’s account with out having to steal their credentials or cookies, successfully getting round MFA safeguards. The third assault sample detected by Microsoft employs compromised credentials, seemingly obtained from a previous occasion, to register their very own phone-based technique to bypass MFA and have interaction in reconnaissance and post-exploitation exercise.

“Following preliminary entry, the actor’s first goal was to rework a short lived compromise right into a persistent foothold,” the Home windows maker stated. “Relatively than relying solely on stolen credentials, the actor enrolled an MFA technique underneath their management, sometimes by registering a brand new telephone quantity, authenticator utility, or software-based one-time password (OTP) token.”

A bonus this actor-controlled second issue presents is that it permits the risk actor to sign-in into the sufferer’s company account with out their participation and keep continued entry together with unrevoked classes or legitimate credentials. The assorted actions the risk actor can take upon establishing MFA persistence are as follows –

• Conduct intensive inside reconnaissance utilizing the Graph API and stock customers, teams, permissions, sources, and accessible content material throughout the tenant utilizing the compromised identification.
• Examine roles and high-value accounts and repair identities for privilege escalation.
• Enumerate mailbox messages, folders, and attachment metadata for intelligence assortment.
• Conduct high-volume entry and obtain exercise aimed toward SharePoint On-line and OneDrive for Enterprise, and even Microsoft Trade On-line in some circumstances.
• Interact in sustained information exfiltration that lasts from a number of hours to a number of days relying on the amount of recordsdata and electronic mail content material harvested from the compromised person.
• Intentionally rotate infrastructure throughout the assault lifecycle and use separate IP addresses for authentication, reconnaissance, and exfiltration actions in order to subvert network-based indicators.

“The assault underscores a vital detection problem: Microsoft Graph abuse not often seems suspicious when seen by way of a single API name,” Microsoft stated. “This assault serves as a robust instance of why Graph exercise have to be assessed holistically, with emphasis on behavioral development and cross-event correlation moderately than particular person API requests in isolation.”