Skip to navigation

Threat reportAPTTL-2026-6406

Russian FSB-linked COLD RELIC (Star Blizzard/COLDRIVER) impersonates employee of hacked US firm to breach second US organization

highACTIVE

Google Threat Intelligence Group reports COLD RELIC, a Russian state-sponsored espionage group assessed by the US DOJ to operate within the FSB, compromised a US organization's Google Workspace tenant in late 2025 using stolen Super Admin credentials and MFA push fatigue.

Key facts

IdentifiersTL-2026-6406
Also known asCOLD RELIC impersonation campaign
SeverityHIGH
First disclosed
Affected productsGoogle Google Workspace; Microsoft Microsoft Entra ID (federated SSO IdP for Google Workspace)
Weaknesses (CWE)CWE-308, CWE-798
AttributionCOLD RELIC, Star Blizzard, Cold River (high confidence)
Nation-state nexusRussia
ToolsProton Mail, Thunderbird
MITRE ATT&CK techniquesT1078.004 Valid Accounts: Cloud Accounts, T1087.004 Account Discovery: Cloud Account, T1090.002 Proxy: External Proxy, T1098.005 Account Manipulation: Device Registration, T1114.002 Email Collection: Remote Email Collection, T1119 Automated Collection, T1528 Steal Application Access Token, T1530 Data from Cloud Storage, T1552.001 Unsecured Credentials: Credentials In Files, T1556.006 Modify Authentication Process: Multi-Factor Authentication, T1566.002 Phishing: Spearphishing Link, T1567.002 Exfiltration Over Web Service: Exfiltration to Cloud Storage and 4 more in the attack flow
MotivationESPIONAGE
Target sectorsgovernment administration, defense, ngo, think-tank, technology
Target regionsNorth America, Europe
Detection rules9 (1 shown free)
Sources8 cited sources

Detection coverage

As of 2026-11-19, Threadlinqs Intelligence publishes 9 detection rule(s) for TL-2026-6406 across Splunk SPL, Microsoft KQL and Sigma, covering 9 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: COLD RELIC Workspace Takeout, Alert Deletion and Credential File Hunting — Post-Exploitation

Correlates Google Workspace Admin and Drive events GTIG ties to COLD RELIC: CUSTOMER_TAKEOUT_CREATED, Batch Delete Operation Performed on Alert Center, User List Download, Investigation Query, searches for .pst/password/cred files and drive sharing to freemail. Fires on any Takeout or two distinct stages by the same actor (TL-2026-6406).

Splunk SPL

index=gws (sourcetype="gws:reports:admin" OR sourcetype="gws:reports:drive") earliest=-24h
| eval ev=coalesce('events{}.name', event_name, name), actor=lower(coalesce('actor.email', actor_email, user))
| eval stage=case(
    ev="CUSTOMER_TAKEOUT_CREATED", "takeout_T1119",
    match(ev,"(?i)batch.?delete"), "alert_center_deletion_T1562.008",
    match(ev,"(?i)user.?list.?download"), "user_list_download_T1087.004",
    match(ev,"(?i)investigation.?query"), "log_visibility_recon_T1562.008",
    match(ev,"(?i)search") AND match(_raw,"(?i)([.]pst|[.]bak|password|cred)"), "credential_file_search_T1552.001",
    match(_raw,"shared_externally") AND match(_raw,"(?i)@gmail[.]com|@proton[.]me|@protonmail[.]com"), "external_drive_share_T1567.002",
    true(), null())
| where isnotnull(stage)
| stats dc(stage) AS stage_count values(stage) AS stages count AS events min(_time) AS first_seen max(_time) AS last_seen by actor
| where stage_count>=2 OR like(mvjoin(stages,","), "%takeout_T1119%")
| convert ctime(first_seen) ctime(last_seen)
| eval detection="COLD RELIC TL-2026-6406 Workspace Super Admin collection and anti-forensics"

Microsoft KQL

let freemail = dynamic(["@gmail.com", "@proton.me", "@protonmail.com"]);
OfficeActivity
| where TimeGenerated > ago(24h)
| where OfficeWorkload in ("SharePoint", "OneDrive")
| where Operation in ("FileDownloaded", "FileSyncDownloadedFull", "SharingSet", "AnonymousLinkCreated", "SearchQueryPerformed")
| where SourceFileName endswith ".pst" or OfficeObjectId has_any (".pst", ".bak", "password", "roster") or TargetUserOrGroupName has_any (freemail)
| summarize Events = count(), Files = dcount(OfficeObjectId), Ops = make_set(Operation, 8), FirstSeen = min(TimeGenerated), LastSeen = max(TimeGenerated) by UserId, ClientIP
| where Events >= 25 or Files >= 20
| project FirstSeen, LastSeen, UserId, ClientIP, Events, Files, Ops

Sigma

title: COLD RELIC Workspace Customer Takeout and Alert Center Tampering
id: 6406a7a0-1002-4a5e-9c10-000000006402
status: experimental
description: Detects Google Workspace Admin log events GTIG attributes to COLD RELIC (TL-2026-6406) - CUSTOMER_TAKEOUT_CREATED domain-wide exports, Alert Center batch deletion to remove evidence, employee User List Download, and Investigation Query used to gauge defender visibility.
references:
    - https://cloud.google.com/blog/topics/threat-intelligence/cold-relic-a-case-study-in-impersonation-and-compromise
    - https://www.cisa.gov/news-events/cybersecurity-advisories/aa23-341a
author: Threadlinqs Team
date: 2026/11/19
tags:
    - attack.collection
    - attack.t1119
    - attack.defense-evasion
    - attack.t1562.008
    - attack.discovery
    - attack.t1087.004
    - attack.t1530
logsource:
    product: google_workspace
    service: google_workspace.admin
detection:
    selection:
        eventName:
            - 'CUSTOMER_TAKEOUT_CREATED'
            - 'Batch Delete Operation Performed'
            - 'User List Download'
            - 'Investigation Query'
    condition: selection
falsepositives:
    - Approved eDiscovery or data-export projects run by Workspace admins
    - Routine Alert Center housekeeping and directory audits by the security team
level: high

Known false positives:

  • Authorised Google Takeout or eDiscovery exports run by Workspace admins
  • Security staff clearing resolved Alert Center notifications
  • Admins performing directory audits

Tuning notes: Allowlist named admins and change-ticket windows for Takeout. Treat CUSTOMER_TAKEOUT_CREATED or Batch Delete of government-backed attack alerts as high priority regardless of stage count. Adjust KQL volume thresholds to tenant norms.

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)
  • COLD RELIC MFA Fatigue Then Sign-in From Residential Proxy ASN (Entra ID) — Exploitation · unlock with TL Creds
  • COLD RELIC Rogue MFA Method Registration After Suspect Sign-in (Entra ID) — Post-Exploitation · unlock with TL Creds
  • COLD RELIC Mail-Scope Consent to Thunderbird or Proton Mail (Entra ID) — C2/Lateral · unlock with TL Creds
  • COLD RELIC Compromised-Site Redirector With file/ref Parameter — Exploitation · unlock with TL Creds
  • COLD RELIC Workspace Customer Takeout and Alert Center Tampering — Post-Exploitation · unlock with TL Creds
  • COLD RELIC Mail-Scope OAuth Grant to Thunderbird or Proton Mail — C2/Lateral · unlock with TL Creds
  • COLD RELIC MFA Push Fatigue Followed by Successful Entra ID Sign-in — Exploitation · unlock with TL Creds
  • COLD RELIC Unapproved OAuth Mail Sync via Thunderbird or Proton Mail — C2/Lateral · unlock with TL Creds

Indicators of compromise

As of 2026-11-19, Threadlinqs Intelligence tracks 9 indicators of compromise for TL-2026-6406.

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

Technical analysis

It then impersonated that organization's employees over an encrypted messaging app to breach a second US organization in 2026 via credential and Microsoft Entra ID token phishing. Persistence relied on rogue MFA devices, passkeys and MFA backup codes.

COLD RELIC (formerly tracked by Google as COLDRIVER and UNC4057; also Star Blizzard, Callisto and Gossamer Bear) is a Russian state-sponsored espionage group that the US DOJ assessed operates within the FSB. GTIG has tracked and disrupted its phishing since 2015. It targets government officials, defense contractors and NGOs in Europe and North America, and GTIG reported new phishing operations as recently as November 2026. The GTIG case study (published 2026-11-17) shows how a single cloud compromise becomes a springboard for compromising third parties: names, email addresses, email archives, contacts and relationship context stolen from one victim are used to build authentic pretexts against downstream targets.

Case study 1 (late 2025, unnamed US Org 1): the actor used a residential proxy geolocated close to the victim together with stolen Super Admin credentials and an MFA push-fatigue attack until the user accepted a prompt. It registered a device model not previously seen in the tenant and persisted with passkeys, MFA backup codes and rogue MFA devices; the passkeys survived password and session resets and bypassed device registration controls. The actor mass-shared the entire cloud storage drive to an actor-operated email account, downloaded files directly, exported full employee lists and started a domain-wide Google Workspace Customer Takeout export (logged as CUSTOMER_TAKEOUT_CREATED) using Super Admin privileges. It queried Admin security investigation and email audit logs to gauge defender visibility, and viewed, modified and deleted Alert Center notifications to remove evidence of the initial intrusion. Over several months it synced inboxes via OAuth to external mail clients, re-authenticated with unrevoked passkeys, and targeted documents on credentials, email archives and personnel information. After removal from Org 1, the actor used email accounts impersonating Org 1 employees to send renewed social engineering to internal staff and associates.

Case study 2 (2026, unnamed US Org 2): the actor impersonated an Org 1 employee over an encrypted messaging app and shared phishing links during the conversation. Phishing links began on a compromised website that redirected to the real phishing domain, hiding it and letting operators fingerprint visitors; link parameters (for example ?file=<previous_victim_name> or ?ref=<lure_name>) tied the link to a prior victim or lure. The pages harvested credentials and authentication tokens for Microsoft Entra ID, the federated SSO identity provider for Google Workspace, and redirected victims to relevant document lures. The actor then registered a new MFA device in the SSO solution and bulk-downloaded files from the user's cloud drive, searching for credentials, personnel information, .pst files, rosters, contact lists, resumes and event schedules to support future spear-phishing. The messaging-app profile was later renamed to a new persona, likely for further targeting of the same professional circles. A COLD RELIC redirector URL referencing Org 1 was submitted to VirusTotal in June 2026.

No CVEs, malware families, file hashes, domains or IPs are published for this campaign; the intrusions rely on valid accounts, MFA abuse and social engineering. GTIG names AS5650 and AS20115 as example residential-proxy ASN ranges to restrict for admin access and lists Google Workspace log events for hunting. GTIG's mitigations: phishing-resistant FIDO2/WebAuthn MFA, monitoring passkey and device enrollments, out-of-band approval of new MFA devices, revoking all sessions, refresh tokens, OAuth grants, passkeys, backup codes and MFA devices (not just resetting passwords) in both Google Workspace and Entra ID, auditing OAuth grants to clients such as Thunderbird and Proton Mail, and DLP on cloud storage.

MITRE ATT&CK techniques by tactic:

Initial Access

T1078.004 Valid Accounts: Cloud Accounts; T1566.002 Phishing: Spearphishing Link

Discovery

T1087.004 Account Discovery: Cloud Account

Command and Control

T1090.002 Proxy: External Proxy; T1665 Hide Infrastructure

Persistence

T1098.005 Account Manipulation: Device Registration; T1556.006 Modify Authentication Process: Multi-Factor Authentication

Collection

T1114.002 Email Collection: Remote Email Collection; T1119 Automated Collection; T1530 Data from Cloud Storage

Credential Access

T1528 Steal Application Access Token; T1552.001 Unsecured Credentials: Credentials In Files; T1621 Multi-Factor Authentication Request Generation

Exfiltration

T1567.002 Exfiltration Over Web Service: Exfiltration to Cloud Storage

Defense Evasion

T1684.001 Impersonation

defense-impairment

T1685.002 Disable or Modify Cloud Log

Remediation

Immediate actions

  • Revoke all active browser sessions, refresh tokens and third-party OAuth grants in Google Workspace and Microsoft Entra ID alongside any password reset
  • Remove unauthorized passkeys, MFA backup codes and rogue MFA devices on admin and compromised accounts
  • Hunt for CUSTOMER_TAKEOUT_CREATED, Batch Delete Operation Performed, User List Download, Investigation Query, Passkey enrolled and Device Registered events in Google Workspace Admin/User logs
  • Audit OAuth grants of the https://mail.google.com/ scope to unapproved clients such as Thunderbird or Proton Mail

Workarounds

  • Restrict admin access from consumer ISPs, VPNs and residential proxy ASNs (examples: AS5650, AS20115) via conditional access
  • Disable Takeout for users at the organization level in Google Workspace

Longer-term hardening

  • Enforce phishing-resistant MFA (FIDO2/WebAuthn hardware keys) and consider Google Advanced Protection Program for high-risk users
  • Require existing-admin or out-of-band approval for new MFA devices and passkeys
  • Enrich authentication logs with residential proxy and VPN IP reputation; alert on sudden residential ASN shifts for privileged accounts
  • Deploy DLP on Google Drive, OneDrive and SharePoint for sensitive terms (password, roster, .pst) and mass external sharing
  • Stop storing credentials in cloud storage; verify out-of-band any contact from a partner organization that was recently breached

Affected and fixed versions

  • Google — Google Workspace
    Vulnerable versions: Tenants without phishing-resistant MFA, session revocation or Takeout/OAuth restrictions
  • Microsoft — Microsoft Entra ID (federated SSO IdP for Google Workspace)
    Vulnerable versions: Tenants without phishing-resistant MFA or token-protection controls

Timeline

  • GTIG begins tracking and disrupting the group's phishing campaigns (year-level precision only).
  • CISA, NCSC-UK and partners publish advisory AA23-341A on Star Blizzard worldwide spear-phishing; DOJ/Treasury actions name FSB officer Ruslan Peretyatko and Andrey Korinets.
  • DOJ unseals warrant seizing 41 domains used by the Callisto Group (FSB Center 18); Microsoft's civil action takes 66 more Star Blizzard domains.
  • Microsoft observes Star Blizzard pivoting to WhatsApp group-invite lures in mid-November 2024 (day approximate).
  • Late 2025 (month approximate): COLD RELIC compromises US Org 1 with stolen Super Admin credentials via residential proxy and MFA push fatigue, registers rogue MFA devices/passkeys, shares the entire Drive externally and starts a Customer Takeout export.
  • 2026 (date approximate): actor retains access for several months via passkeys, backup codes and OAuth mail sync, and deletes Alert Center notifications; after removal, uses accounts impersonating Org 1 employees.
  • 2026 (date approximate): actor impersonates an Org 1 employee over an encrypted messaging app, phishes Entra ID credentials and tokens from US Org 2, registers a new MFA device and bulk-downloads Drive files.
  • June 2026: a COLD RELIC redirector URL referencing Org 1 is submitted to VirusTotal.
  • November 2026: GTIG observes new COLD RELIC phishing operations.
  • GTIG publishes 'COLD RELIC: A Case Study in Impersonation and Compromise'.
  • SecurityWeek reports on the GTIG findings.

Sources cited

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