Google: Russia's COLD RELIC Reuses Stolen Contact Lists to Impersonate Colleagues and Breach Google Workspace and Entra ID Accounts, Still Active

Google's Threat Intelligence Group (GTIG) has published two chained COLD RELIC intrusion cases. It says the group used data stolen in one breach to pose as trusted contacts to the next target. Google also offers detection and hardening advice and says the group remains active.

News date: About 7 min read

Share on LINE
Contents
  1. 1. Who is COLD RELIC, and what did Google disclose?
  2. 2. How do the two intrusions differ from past practice?
  3. 3. What did the attackers do after getting in?
  4. 4. How can organizations and administrators harden?
  5. 5. FAQ

Google's Threat Intelligence Group (GTIG) and Mandiant have published a case study on COLD RELIC, a Russian cyber-espionage group. This is not a product update but a public threat report. Google says the group remains a threat and that new phishing campaigns were still being observed in November 2026. Two points matter for ordinary organizations. First, a message impersonating an acquaintance may use real data from a previous victim. Second, resetting a password may not drive the attacker out.

Who is COLD RELIC, and what did Google disclose?

According to Google, COLD RELIC was previously tracked as COLDRIVER and UNC4057. It is also known as Star Blizzard, Callisto and Gossamer Bear. Google cites a US Department of Justice assessment that the group operates within Russia's Federal Security Service (FSB). Google says the group targets government officials, defense contractors and non-governmental organizations in Europe and North America.

Google says Mandiant handled a series of related intrusions over the past year. The most recent occurred in 2026. In that intrusion, attackers used data, employee contact lists and identities stolen from another US organization in late 2025 to target a second organization. GTIG also says it has tracked and disrupted the group's phishing campaigns since 2015. It sends government-backed attack alerts when it detects the group targeting Google accounts.

How do the two intrusions differ from past practice?

Google says the new insight is that the group turned one cloud breach into a springboard to compromise other organizations. The attackers used stolen emails and data to build credible pretexts, then posed as acquaintances to approach the next target. In an older January 2024 post, Google noted that the group often used impersonation accounts to build rapport before sending phishing links. That post also said the group had used a backdoor called SPICA.

A few terms first. Multi-factor authentication (MFA) is an extra confirmation at login beyond the password, such as tapping approve on a phone. In a push-fatigue attack, an attacker who already holds the password repeatedly triggers login prompts until the user taps approve. Google says the victim in case one accepted an MFA request. Single sign-on (SSO) means using one account to log in to multiple services. "Federated" means another company's sign-in system handles that login. Microsoft Entra ID is Microsoft's account sign-in service. A Super Admin is the highest-level administrator account in Google Workspace. A residential proxy relays connections through ordinary home internet connections. Google says the attackers chose residential proxies near the victim's location so that logins looked normal. A passkey is a login credential that needs no password. OAuth is an authorization mechanism that lets third-party apps access an account.

Comparison of the two cases based on Google's report
AspectCase 1 (late 2025)Case 2 (2026)
Initial accessPer Google, residential proxies, stolen Super Admin credentials and an MFA push-fatigue attackPer Google, posing as an employee of the case-1 organization and sending phishing links via an encrypted messaging app
Affected environmentGoogle Workspace cloud productivity environmentAccounts using Microsoft Entra ID as federated SSO for Workspace
Persistence (keeping long-term access)Google says passkeys, MFA backup codes and rogue MFA devices were usedGoogle says a new MFA device was registered
Data actionsGoogle says the contents of the entire organization's cloud storage drive were downloaded, employee lists exported, and Customer Takeout (an admin tool for exporting organization-wide data) startedGoogle says files in that user's cloud drive were downloaded in bulk

What did the attackers do after getting in?

Google says that in case one the attackers queried the admin console audit logs to see what defenders could see. They also viewed, modified and deleted Alert Center notifications to remove evidence. For months afterward they kept access, for example by syncing the inbox to an external mail client via OAuth. They also logged back in using unrevoked passkeys, which remained valid after password and session resets.

Google adds that the attackers searched for credentials, personnel information and .pst mail archives. They also searched for rosters, contact lists, résumés and event schedules. That material could be used for later spear phishing, meaning phishing tailored to specific targets. After the victim organization removed them, the attackers used email accounts impersonating its employees. They used those accounts to approach internal colleagues and associates in the same professional circles.

How can organizations and administrators harden?

  • Authentication: Google recommends replacing SMS, voice and push verification with FIDO2/WebAuthn hardware keys. It also recommends auditing passkey registrations and unusual device enrollments.
  • MFA enrollment: adding a new MFA device should require approval from an existing administrator or out-of-band identity verification (confirming identity through a separate, independent channel).
  • OAuth: continuously review third-party grants in Google Workspace and Microsoft 365. Watch for unexpected desktop mail clients or external mail services.
  • Network: enrich sign-in logs with residential proxy and VPN IP reputation feeds. Restrict admin access from consumer ISPs, VPNs and known residential proxy networks.
  • Data loss prevention (DLP): Google recommends DLP policies that alert on files containing sensitive terms or restrict mass external sharing.
  • Advanced protection: Google says organizations and individuals may consider enrolling in the Advanced Protection Program.

For detection, Google lists several Workspace log signals: - a successful login following a burst of repeated MFA prompts - creation of a Customer Takeout - bulk downloads or external sharing in a short time - new passkey or device registrations - batch deletion in Alert Center Google SecOps customers can find related rules in the Mandiant Frontline Threats rule pack, for example "Google Workspace Customer Takeout Initiated".

FAQ

Is this a new software vulnerability?

According to Google's report, the cases describe social engineering, credential and MFA abuse, and cloud identity persistence techniques. The report does not mention a software vulnerability.

Why are impersonation messages so hard to spot?

Google says the attackers built realistic pretexts from contact lists and emails stolen from the previous victim. They then posed as acquaintances in the same professional circle. Their phishing links first go to a compromised website, which redirects to the actual phishing domain. In case two, the links then led the victim to document lures relevant to the conversation.

Is this still ongoing?

Google says COLD RELIC remains an active threat, and GTIG observed new phishing campaigns as recently as November 2026.

Do only organizations using Google Workspace need to pay attention?

No. Google's case two involves Microsoft Entra ID accounts. Google's recommendations also cover revoking sessions and OAuth grants in Microsoft Entra ID and Microsoft 365, and they mention cross-cloud DLP.

What should I do if I receive a suspicious message from an acquaintance?

The report gives no step-by-step guidance for individuals. Its core observation is that attackers borrow trusted relationships. Its recommendations are aimed mainly at organizations, such as moving to FIDO2/WebAuthn hardware keys and auditing passkeys and OAuth grants.

Browse the latest news in this topic

Found this guide helpful? Share it with someone who needs it

Share on LINE

Sources

Compiled from the references and verification dates listed above; fares, schedules, product features, and terms can change, so please confirm the latest details on the relevant official website before travelling, booking, or purchasing.

Lifestyle