Google says Cold Relic turned stolen data into new intrusions

Google says Cold Relic turned stolen data into new intrusions

Add CyberWorldOpsto your preferred sources on Google
Google says Cold Relic turned stolen data into new intrusions
APT

Illustrative image generated with AI

Russian espionage group Cold Relic used information stolen from one US organization to help compromise another, Google’s Mandiant and Threat Intelligence Group said in a November 17 report. The linked intrusions show how stolen staff directories, emails and identities can support convincing approaches to trusted contacts—and how attacker-enrolled authentication methods can preserve access after password resets.

Mandiant investigated both the late-2025 compromise and the subsequent intrusion at a separate US organization in 2026. The second organization used Microsoft Entra ID to provide federated single sign-on, allowing users to access Google Workspace through their Microsoft identity. GTIG observed evidence of new Cold Relic phishing operations as recently as November 2026.

The group, formerly tracked as COLDRIVER and UNC4057 and also known as Star Blizzard, Callisto and Gossamer Bear, has successfully targeted government officials, defense contractors and nongovernmental organizations across Europe and North America. Google describes it as Russian state-sponsored and cites the US Department of Justice’s assessment that it operates within Russia’s Federal Security Service, or FSB. GTIG has tracked and disrupted its phishing campaigns since 2015.

From an administrator account to a second organization

In the late-2025 intrusion, Cold Relic signed in with stolen credentials belonging to a Super Admin account, which carried extensive administrative privileges. It combined those credentials with a multi-factor authentication (MFA) fatigue attack: repeated approval requests that ended when the victim accepted one.

The attackers connected through a residential proxy, which routed their traffic through a consumer internet connection near the victim’s location. Google said the group deliberately chooses nearby proxies to evade alerts based on implausible travel between successive sign-ins.

After entering the organization’s cloud environment, Cold Relic registered a previously unseen device model. It also used passkeys, MFA backup codes and unauthorized authentication devices to retain access.

The Hacker News, reporting additional details from Google’s account, said the attackers used Super Admin privileges to download the organization’s entire cloud drive through direct downloads and mass sharing to an attacker-controlled email account. They also exported employee lists and domain-wide data through Google Takeout.

The same account described attackers querying security investigation and email audit logs to assess defensive visibility, then viewing, modifying and deleting Alert Center notifications to remove evidence. They maintained access for several months, synchronized email to external clients through OAuth—an authorization mechanism that grants applications account access—and sought credentials, email archives, personnel information and other documents. Unrevoked attacker-enrolled passkeys survived password and session resets and bypassed device registration controls.

Even after their removal, the attackers used email accounts impersonating employees of the first organization to approach its staff and professional contacts.

That relationship knowledge helped underpin the next intrusion. Mandiant found that Cold Relic approached a target at the second organization through an encrypted messaging app while impersonating an employee of the first. The lure offered a document about a technical issue; phishing links sought Entra ID credentials and authentication tokens, which can grant access to authenticated sessions.

Google said phishing URL parameters and impersonation pretexts indicated that the attackers likely used contact and relationship information from the first victim. GTIG also correlated a redirector URL submitted to VirusTotal in June 2026 whose query parameter referenced that organization.

Cold Relic commonly places a compromised website between the recipient and its phishing site. That intermediary hides the final destination and lets the attackers assess visitors before deciding whether to redirect them. Link parameters can refer to a previous victim or the lure’s subject.

After compromising the second organization’s user account, Cold Relic registered another MFA device in the single sign-on system and bulk-downloaded files available through that user’s cloud drive, Google said. The messaging account subsequently changed its profile name to another persona, which researchers suspected was intended for further targeting within the same professional network.

Removing access requires more than a password reset

Google found that credentials and other secrets stored in cloud files amplified some intrusions, allowing attackers to reach additional services and personal accounts typically outside employers’ monitoring. In at least one case, an organization reset passwords but failed to revoke existing sessions, leaving Cold Relic connected.

For administrators handling suspected compromise, Google recommends revoking active browser sessions, refresh tokens and third-party OAuth permissions across both Google Workspace and Microsoft Entra ID. They should also remove unauthorized passkeys, backup codes and MFA devices. Password changes alone are insufficient.

For prevention, Google recommends phishing-resistant MFA using FIDO2/WebAuthn hardware security keys instead of SMS, voice or push approvals. These keys bind authentication to the legitimate site, resisting intermediary phishing proxies and approval-fatigue attacks. New authentication devices should require approval from an existing administrator or separate identity verification. Organizations should audit passkey registrations and device enrollment, especially on privileged accounts; organizations and individuals can also consider Google’s Advanced Protection Program.

Google’s detection guidance calls for security teams to watch for:

  • Repeated failed sign-ins or MFA challenges followed by success from the same IP address; unexpected device or passkey enrollment; and backup-code authentication from unusual networks.
  • New OAuth permissions granting broad mailbox access to unapproved clients or external mail services, such as Thunderbird or Proton Mail; spikes in downloads or external sharing; organization-wide Takeout exports; and employee-directory exports from unknown devices.
  • Security-alert deletion or modification, unexpected audit-log queries, and searches for credential or archive terms such as .pst, .bak, password and cred.

Administrators should continuously review third-party application permissions in Google Workspace and Microsoft 365. Authentication monitoring should incorporate residential-proxy and VPN reputation data and flag sudden changes in the internet-provider network used by privileged accounts. Google also recommends conditional-access policies to block or detect known proxy connections and restrictions on administrative access from consumer providers, VPNs and known residential-proxy networks, including AS5650 and AS20115.

To reduce opportunities for further compromise and theft, Google recommends data loss prevention rules across Google Drive, OneDrive and SharePoint. These can warn or block uploads and external sharing of files containing secrets, flag terms such as passwords, roster or .pst, and restrict mass external file-sharing.

Read next

Sources

Back to home

Latest Cybersecurity News

All cybersecurity news →