Skip to navigation

Threat reportAPTTL-2026-2167

Russian Cyber Espionage Infrastructure Uses Evilginx and OAuth Phishing to Steal Accounts

highACTIVE

Three Russian-nexus clusters — UNC6293 and UNC7005 (subclusters of ICE RELIC/APT29/Cozy Bear/Midnight Blizzard, UNC7005 also tracked as Storm-2945) and the independently-attributed UNC5976 — are compromising government, academic, aerospace, defense, and think-tank accounts across Europe, the US, Ukraine, and Armenia by abusing legitimate authentication features (app-specific passwords, OAuth consent, Microsoft/WhatsApp device-code flows) and Evilginx adversary-in-the-middle reverse-proxy infrastructure rather than exploiting software vulnerabilities.

Key facts

IdentifiersTL-2026-2167
Also known asCaptiveCrunch
SeverityHIGH
Affected productsGoogle Google Workspace OAuth / Application-Specific Passwords; Microsoft Microsoft Entra ID device code authentication flow / Microsoft 365; Meta WhatsApp device linking
AttributionUNC6293, UNC7005, UNC5976 (medium confidence)
Nation-state nexusRussia
MalwareAMOS, Evilginx, HEADRUSH, Vidar
Toolsevilginx2 - S9003
MITRE ATT&CK techniquesT1059.007 JavaScript, T1071.001 Web Protocols, T1090.002 External Proxy, T1111 Multi-Factor Authentication Interception, T1123 Audio Capture, T1125 Video Capture, T1204.001 Malicious Link, T1204.002 Malicious File, T1497.001 System Checks, T1528 Steal Application Access Token, T1539 Steal Web Session Cookie, T1550.001 Application Access Token and 7 more in the attack flow
MotivationESPIONAGE
Target sectorsgovernment administration, academia, aerospace, defense, think tanks, ngo, diplomatic, hospitality
Target regionsEurope, united states of america, ukraine, armenia
Detection rules9 (1 shown free)
Sources8 cited sources

Detection coverage

As of 2026-08-27, Threadlinqs Intelligence publishes 9 detection rule(s) for TL-2026-2167 across Splunk SPL, Microsoft KQL and Sigma, covering 30 indicator(s) of compromise. The whole corpus is readable without an account. Every threat comes with one complete detection in Splunk SPL, Microsoft KQL and Sigma, free for everyone; a free account adds daily TL Creds to unlock the rest. Paid plans add a larger daily allowance, correlation and the MCP server. Threadlinqs MCP server · View plans.

Free detection: Evilginx AiTM / OAuth-Lure Phishing Domain Contact — stateaffairs.us Cluster

Detects web/proxy traffic to the UNC6293/UNC7005/UNC5976 Evilginx2 reverse-proxy and OAuth-lure phishing domains (stateaffairs.us, foreignrelations.us, dosportal.app, internationalaffairsportal.us, my-invite.org, owa-ms365.com/m365-owa.com/ms365-device.com, foc-share.com, verify-drive.com family, chamber-ua.org, mail.kiis.co.uk) or their known hosting IPs. These sites transparently relay victims to real State Department / Microsoft OWA pages while an Evilginx-style reverse proxy captures credentials, session cookies, and OAuth 'verification codes' in transit.

  • Languages: Splunk SPL, Microsoft KQL, Sigma
  • ATT&CK: T1566.002, T1583.001, T1557, T1090.002
  • Data sources: proxy, dns, device_network_events, email_url_info
  • Severity: critical
  • Confidence: high

Splunk SPL

index=proxy (sourcetype=corporate_proxy OR sourcetype=zscaler OR sourcetype="bluecoat:proxysg" OR sourcetype=pan:threat)
| eval phish_domain=case(
    like(dest_domain, "%stateaffairs.us%"), "stateaffairs.us",
    like(dest_domain, "%foreignrelations.us%"), "foreignrelations.us",
    like(dest_domain, "%dosportal.app%"), "dosportal.app",
    like(dest_domain, "%internationalaffairsportal.us%"), "internationalaffairsportal.us",
    like(dest_domain, "%my-invite.org%"), "my-invite.org",
    like(dest_domain, "%owa-ms365.com%") OR like(dest_domain, "%m365-owa.com%") OR like(dest_domain, "%ms365-device.com%"), "owa-spoof-cluster",
    like(dest_domain, "%foc-share.com%"), "foc-share.com",
    like(dest_domain, "%verify-drive.com%"), "verify-drive-cluster",
    like(dest_domain, "%chamber-ua.org%"), "chamber-ua.org",
    like(dest_domain, "%kiis.co.uk%"), "mail.kiis.co.uk",
    dest_ip="151.236.15.213" OR dest_ip="185.158.250.155" OR dest_ip="93.127.160.28" OR dest_ip="104.194.159.150", "known-hosting-ip",
    1=1, null())
| where isnotnull(phish_domain)
// AiTM reverse proxies often relay a real login POST and set a session cookie header shortly after
| stats count min(_time) as first_seen max(_time) as last_seen values(http_method) as methods by src_user, src_ip, dest_domain, dest_ip, phish_domain
| where count > 0
| sort -count

Microsoft KQL

let PhishDomains = dynamic(["stateaffairs.us","foreignrelations.us","dosportal.app","internationalaffairsportal.us","my-invite.org","owa-ms365.com","m365-owa.com","ms365-device.com","foc-share.com","verify-drive.com","chamber-ua.org","mail.kiis.co.uk"]);
let PhishIPs = dynamic(["151.236.15.213","185.158.250.155","93.127.160.28","104.194.159.150"]);
DeviceNetworkEvents
| where Timestamp > ago(24h)
| where RemoteUrl has_any (PhishDomains) or RemoteIP in (PhishIPs)
| project Timestamp, DeviceName, InitiatingProcessAccountName, RemoteUrl, RemoteIP, InitiatingProcessFileName
| join kind=leftouter (
    EmailUrlInfo
    | where Timestamp > ago(24h)
    | where Url has_any (PhishDomains)
    | project EmailUrl_Timestamp = Timestamp, RecipientEmailAddress, Url
  ) on $left.RemoteUrl == $right.Url
| summarize Hits = count(), FirstSeen = min(Timestamp), LastSeen = max(Timestamp) by DeviceName, InitiatingProcessAccountName, RemoteUrl, RemoteIP
| order by Hits desc

Sigma

title: Evilginx AiTM / OAuth-Lure Phishing Domain Contact (UNC6293/UNC7005/UNC5976)
id: a1e4c2d0-6b3a-4f1e-9c7d-2b8f5a3e1d40
status: experimental
description: Detects proxy/DNS traffic to Evilginx2 reverse-proxy and OAuth-verification-code lure domains used by Russian-nexus clusters UNC6293, UNC7005 (Storm-2945), and UNC5976 to relay victims to real login pages while harvesting credentials, session cookies, and OAuth grants.
references:
    - https://cloud.google.com/blog/topics/threat-intelligence/distinct-clusters-target-individuals-of-interest-to-russia
    - https://attack.mitre.org/software/S9003/
author: Threadlinqs Team
date: 2026-08-27
tags:
    - attack.resource-development
    - attack.t1583.001
    - attack.credential-access
    - attack.t1557
    - attack.command-and-control
    - attack.t1090.002
    - attack.tool.evilginx2
    - attack.g.unc6293
logsource:
    category: proxy
    product: windows
detection:
    selection_domain:
        c-uri|contains:
            - 'stateaffairs.us'
            - 'foreignrelations.us'
            - 'dosportal.app'
            - 'internationalaffairsportal.us'
            - 'my-invite.org'
            - 'owa-ms365.com'
            - 'm365-owa.com'
            - 'ms365-device.com'
            - 'foc-share.com'
            - 'verify-drive.com'
            - 'chamber-ua.org'
            - 'mail.kiis.co.uk'
    selection_ip:
        dst_ip:
            - '151.236.15.213'
            - '185.158.250.155'
            - '93.127.160.28'
            - '104.194.159.150'
    condition: 1 of selection_*
falsepositives:
    - Security researchers or automated threat-intel crawlers deliberately visiting known IOCs for enrichment
level: high

Known false positives:

  • Security researchers, sandboxes, or automated threat-intel crawlers deliberately visiting the listed domains/IPs for enrichment or takedown work

Tuning notes: Maintain the domain/IP list as a lookup table refreshed from the harness's threat_iocs feed so newly registered spoof domains (the actors rotate registrars/registrant emails such as given956@2200freefonts.com and chikolimdrid@gmail.com) are picked up automatically; add TLS JA3/JA3S fingerprinting on the Evilginx reverse-proxy handshake if available to catch not-yet-listed subdomains of the same infrastructure.

8 more detection rules (Splunk SPL, Microsoft KQL, Sigma) · unlock with TL Creds. Sign up free to unlock the detection rules or query them with the Threadlinqs MCP server (Purple)
  • Device-Code / Embassy-Invite Phishing Link Click — UNC7005 Conference Lure Template · unlock with TL Creds
  • WhatsApp Device-Link / Attacker-Injected JavaScript Call Recording Indicator · unlock with TL Creds
  • Captive-Portal / Hospitality Wi-Fi AiTM Redirect (CaptiveCrunch) to CornFlake RAT / ChocoShell Delivery · unlock with TL Creds
  • DNS Resolution of Evilginx / OAuth-Lure Phishing Infrastructure · unlock with TL Creds
  • HEADRUSH Excel Add-In to HTA Downloader / ENGINELIGHT-CHERRYPIE PowerShell Execution Chain · unlock with TL Creds
  • Outbound Connection to Vidar/HEADRUSH File-Hash-Confirmed Hosts or Community-Corroborated Sibling C2 · unlock with TL Creds
  • Suspicious ASP / OAuth Consent Grant / Device-Code Approval — Identity Provider Abuse · unlock with TL Creds
  • Infostealer / RAT C2 Beaconing — Vidar, ENGINELIGHT, CHERRYPIE, CornFlake RAT (incl. Dead-Drop Resolver Abuse) · unlock with TL Creds

Indicators of compromise

As of 2026-08-27, Threadlinqs Intelligence tracks 30 indicators of compromise for TL-2026-2167.

Sample indicators (3 of 30)

  • IP address: 151.236.15[.]213
  • Domain: stateaffairs[.]us
  • File hash: c66f20f2e39eb2f6a0a4cdbe0d955e5f

Defanged, TLP:CLEAR. Full set of 30 indicators: unlock with TL Creds.

30 indicators of compromise · unlock with TL Creds. Compare plans

Technical analysis

The campaign, active from at least June 2025 through August 2026, bypasses MFA, harvests session cookies/OAuth tokens, and delivers infostealer and RAT payloads (Vidar, Atomic/AMOS, HEADRUSH, CHERRYPIE/ChocoShell, ENGINELIGHT, CornFlake RAT) via credential-theft and malware-as-a-service infrastructure.

Google Threat Intelligence Group (GTIG) publicly linked three distinct but related Russian-nexus clusters conducting large-scale account-compromise operations against individuals and institutions of interest to Russia. UNC6293, first detailed by Google and Citizen Lab in June 2025, is assessed with moderate confidence as an initial-access subcluster of ICE RELIC (the successor tracking name for APT29/Cozy Bear/Midnight Blizzard/Nobelium/BlueBravo). UNC6293 began by impersonating US State Department officials to solicit application-specific passwords (named e.g. 'ms.state.gov') via email and PDF walkthroughs, then evolved by October 2025 to harvesting ASPs entered on attacker phishing sites, and by June 2026 to abusing OAuth by asking targets to relay a post-login 'verification code' after a legitimate sign-in — effectively handing the attacker an OAuth authorization grant. This activity is hosted on Evilginx-style reverse-proxy infrastructure (subdomains of stateaffairs[.]us and related domains such as foreignrelations[.]us, dosportal[.]app, and internationalaffairsportal[.]us) that transparently relays victims to real US State Department pages while capturing credentials, session cookies, and MFA tokens in transit — the classic adversary-in-the-middle (AiTM) pattern documented for the open-source evilginx2 framework.

UNC7005 (aka Storm-2945), first identified in February 2026 and linked by Microsoft to Midnight Blizzard, targets academia, diplomatic personnel, and NGOs in Ukraine, Western Europe, and the US with lower operational security than UNC6293 but a heavier malware and infrastructure footprint. UNC7005 conducts Microsoft Entra ID and WhatsApp device-code phishing: victims are lured via spoofed conference/embassy-invite pages (reusing a template across an 'embassy invite' April 2026 lure and a GLOBSEC Forum 2026 spoof in May 2026) into approving a device-code login, or into linking WhatsApp to an attacker-controlled device under the pretense of joining a secure call — after which attacker-injected JavaScript can silently record audio/video during the resulting 'call' or harvest chat credentials. From May 2026, Microsoft/ReliaQuest/Zscaler documented a related sub-campaign, CaptiveCrunch, in which UNC7005/Storm-2945 manipulated shared hospitality-sector Wi-Fi captive portals worldwide to redirect travelers to Microsoft 365 credential-harvesting and device-code phishing pages, escalating from mid-July 2026 into direct AiTM abuse of the Entra ID device-code flow and delivering CornFlake RAT and ChocoShell (CHERRYPIE). In parallel, UNC7005 ran a malware-as-a-service infostealer campaign (May 2026) delivering Vidar (Windows, Go-based, C2 107.189.18.7) and Atomic/AMOS (macOS) via summit-themed phishing, and a smaller ENGINELIGHT Go-malware operation (April–May 2026, C2 statistic-ms.live) distributed from bounce@chamber-ua.org. By July–August 2026, UNC7005 registered domains spoofing the Finnish Operations Center and Microsoft OWA to target the European defense industry directly with OAuth/device-code phishing, including sending legitimate Microsoft OAuth URLs straight to victims and abusing unverified Google Cloud test-mode projects to capture OAuth tokens.

UNC5976, active since March 2026 and assessed as a separate Russian-nexus cluster (possibly a different intelligence service), focuses on military, aerospace, defense-industrial, NGO, and think-tank targets in Ukraine and Armenia. It automates OAuth token theft using fake file-sharing domains (verify-drive[.]com, drive.google.verify-drive[.]com) paired with attacker-owned Google Cloud projects: victims see an automatic 'Continue with Google' popup, complete a legitimate OAuth login, and are silently redirected through a malicious script that exfiltrates the resulting access token. UNC5976 also distributed HEADRUSH, a malicious Excel add-in leading to an HTA downloader, from a domain masquerading as a Ukrainian research institute, likely targeting a Ukrainian aerospace/imaging company. Following GTIG's disruption of its Google-hosted infrastructure, UNC5976 has migrated to non-Google cloud providers.

Across all three clusters, the unifying tradecraft is abuse of legitimate identity and authentication workflows — app-specific passwords, OAuth consent screens, and device-code/QR-based linking — rather than a software vulnerability, making the activity resistant to traditional patch-based remediation and dependent on identity-layer controls (phishing-resistant MFA, ASP restriction, OAuth-grant auditing, and device-link monitoring) for detection and response.

MITRE ATT&CK techniques by tactic:

Execution

T1059.007 JavaScript; T1204.001 Malicious Link; T1204.002 Malicious File

Command and Control

T1071.001 Web Protocols; T1090.002 External Proxy

Credential Access

T1111 Multi-Factor Authentication Interception; T1528 Steal Application Access Token; T1539 Steal Web Session Cookie; T1557 Adversary-in-the-Middle

Collection

T1123 Audio Capture; T1125 Video Capture

Defense Evasion

T1497.001 System Checks

lateral-movement

T1550.001 Application Access Token

Initial Access

T1566.001 Spearphishing Attachment; T1566.002 Spearphishing Link

Resource Development

T1583.001 Domains; T1583.006 Web Services; T1585.002 Email Accounts; T1588.001 Malware

Remediation

Immediate actions

  • Disable application-specific passwords org-wide, or restrict account security to 'Security Keys only' for at-risk users
  • Block known malicious domains and IPs (stateaffairs.us cluster, UNC7005/UNC5976 infrastructure) at email gateways, DNS, and web proxies
  • Audit and revoke suspicious OAuth app grants and unverified/testing-mode cloud-project consents on Google Workspace and Microsoft 365 tenants
  • Audit and remove unrecognized linked devices on WhatsApp and similar linking-capable platforms

Workarounds

  • Require independent, off-platform verification of conference, embassy, and event invitations before registering or authenticating
  • Disable or closely monitor device-code authentication flows for Microsoft Entra ID tenants where not operationally required
  • Advise traveling staff to use VPN rather than authenticating directly through hotel/conference Wi-Fi captive portals

Longer-term hardening

  • Enforce phishing-resistant MFA (FIDO2/WebAuthn security keys) for diplomats, academics, defense-industry, and think-tank personnel
  • Enroll high-risk individuals in Google Advanced Protection Program / Microsoft equivalent
  • Deploy conditional access policies that block or flag OAuth device-code and app-password authentication paths by default
  • Monitor captive-portal and public Wi-Fi authentication traffic for DNS/HTTP manipulation consistent with CaptiveCrunch-style AiTM redirection

Affected and fixed versions

  • Google — Google Workspace OAuth / Application-Specific Passwords
    Vulnerable versions: legitimate feature abuse, not a software flaw
    Fixed in: N/A - identity/policy controls apply
  • Microsoft — Microsoft Entra ID device code authentication flow / Microsoft 365
    Vulnerable versions: legitimate feature abuse, not a software flaw
    Fixed in: N/A - identity/policy controls apply
  • Meta — WhatsApp device linking
    Vulnerable versions: legitimate feature abuse, not a software flaw
    Fixed in: N/A - identity/policy controls apply

Timeline

  • UNC6293 application-specific-password phishing campaign first publicly detailed by Google and Citizen Lab, impersonating US State Department officials with an 'ms.state.gov' app-password lure.
  • UNC6293 evolves ASP phishing to have targets enter their app password directly into an attacker-controlled phishing website rather than share it via email.
  • Volexity documents UNC6293 diplomatic/security-event-themed phishing lures targeting European security-conference attendees.
  • UNC7005 (Storm-2945) identified as a distinct cluster, beginning high-risk targeting of academia, diplomatic personnel, and NGOs in Ukraine, Western Europe, and the US.
  • UNC5976 begins automated OAuth token-collection operations using fake file-sharing domains paired with attacker-owned Google Cloud projects.
  • UNC5976 distributes the HEADRUSH malicious Excel add-in/HTA downloader from a domain spoofing a Ukrainian research institute; UNC7005 stands up locked-ioc-22 device-code phishing infrastructure resolving to locked-ioc-18.
  • UNC7005 launches WhatsApp device-linking abuse operations (with JavaScript-based audio/video recording of resulting calls) and a summit-themed malware-as-a-service campaign delivering Vidar and Atomic/AMOS infostealers; GLOBSEC Forum 2026 registration page spoofed, reusing the 'embassy invite' device-code phishing template.
  • UNC6293 begins OAuth 'verification code' phishing via locked-ioc-13, impersonating US State Department officials to obtain OAuth authorization grants after a legitimate login.
  • UNC7005 registers Microsoft OWA-themed spoof domains (locked-ioc-23, m365-owa.com, ms365-device.com) escalating the hospitality-network AiTM campaign.
  • ReliaQuest reports DNS/HTTP traffic manipulation on shared hotel and conference-center Wi-Fi captive portals redirecting travelers into Microsoft Entra ID device-code AiTM phishing.
  • Microsoft publicly attributes the CaptiveCrunch captive-portal campaign to Storm-2945/Midnight Blizzard, detailing delivery of CornFlake RAT and ChocoShell (CHERRYPIE).
  • UNC7005 sends Finnish Operations Center-themed spearphishing emails to European defense-industry targets from locked-ioc-24 and related domains, continuing through August 13, 2026.
  • Validin publishes an infrastructure analysis of the locked-ioc-15 Evilginx reverse-proxy phishing cluster tied to the broader campaign.

Sources cited

Previous and next in apt

Drafted by the Threadlinqs research pipeline from 8 sources; published ; last updated —.

Threadlinqs Intelligence — Real-Time Threat Detection Platform

// threat_feed
$ sort --newest
Showing all threats

Live intelligence console

Threat level

Latest Threats