The Russia-linked cyber espionage collective known as Cold Relic—historically tracked under aliases such as COLDRIVER, UNC4057, Callisto, Gossamer Bear, and Star Blizzard—has evolved its tactical playbook. Cybersecurity researchers have observed the advanced threat group executing sophisticated, multi-stage operations by turning a single initial cloud compromise into a high-utility launchpad for subsequent, interconnected intrusions targeting third-party organizations.

Recent investigative findings released by Google’s Mandiant and Threat Intelligence Group (GTIG) shed light on a complex cyber incident investigated in 2026. During this campaign, the adversary cleverly leveraged internal corporate data, staff directories, and compromised identities harvested during a late-2025 intrusion at a separate U.S.-based entity. This methodical pivoting highlights a dangerous evolution in how state-sponsored cyber actors abuse cloud productivity environments, transforming stolen digital footprints into reliable vectors for lateral movement across organizational boundaries.

Cold Relic has long maintained a formidable reputation within the global threat intelligence community. The group possesses a well-documented track record of orchestrating highly targeted phishing campaigns, credential harvesting operations, and bespoke malware deployments. Their primary objective generally revolves around acquiring strategic intelligence of vital interest to the Russian government. Over the years, their operations have successfully compromised high-value targets, including prominent government officials, defense contractors, and various non-governmental organizations operating across North America and Europe.

The mechanics of Cold Relic’s intrusions typically rely on exploiting human trust and modern cloud infrastructure. Once initial access is secured, the threat actors rapidly abuse their footing to register rogue multi-factor authentication devices, passkeys, and MFA backup codes. These measures allow the group to establish long-term persistence within a victim’s network. With persistence secured, the actors pivot to aggressive data collection and exfiltration operations, focusing heavily on corporate email repositories, chat logs, and cloud storage systems. Furthermore, compromised accounts are frequently weaponized to target the victim’s broader professional and social networks.

"Once Cold Relic compromises a target, it uses stolen emails and data to establish authentic pretexts and impersonate contacts when reaching out to subsequent targets," Google explained in its detailed case study on the campaign.

Russian COLD RELIC Hackers Use Stolen Google Workspace Data to Breach Second U.S. Organization

According to Google’s cybersecurity and incident response specialists, the sprawling campaign began in late 2025 when Cold Relic successfully infiltrated the cloud productivity environment of an unspecified U.S.-based organization. Initial access to the corporate network was achieved through a residential proxy network strategically positioned in close geographical proximity to the victim organization, helping the attackers blend in with typical regional traffic patterns and evade initial geographic-based anomaly detection systems.

The breakthrough into the environment relied on a combination of credential theft targeting a Super Admin account and an aggressive multi-factor authentication push notification fatigue attack. By bombarding the targeted administrator with relentless MFA requests, the attackers eventually wore down the victim’s vigilance until an approval was mistakenly granted. Capitalizing on this momentary lapse, the threat actor swiftly registered their own rogue device within the cloud tenant to guarantee ongoing access. Alongside the newly registered device, the actors incorporated passkeys and MFA backup codes as resilient alternative access mechanisms, ensuring that even if primary tokens expired, their foothold remained secure.

With administrative privileges firmly in hand, the threat actor moved quickly to harvest organizational assets. "Cold Relic used the privileges of the compromised Super Admin account to download the contents of the entire organization’s cloud storage drive by both mass sharing files to a Cold Relic-operated email account and downloading files directly," Google detailed in its technical breakdown.

The operational footprint left behind inside the cloud tenant revealed a meticulous effort to conceal their presence and destroy forensic artifacts. While operating within the victim’s environment, Cold Relic personnel accessed the administrative console, queried admin security investigation logs, and reviewed email audit trails to assess the organization’s defensive visibility. Moreover, the actors actively viewed, modified, and deleted Alert Center notifications specifically to remove forensic evidence of their initial intrusion and subsequent movements.

These administrative privileges ultimately facilitated the large-scale export of comprehensive employee lists, corporate directories, and domain-wide data via the built-in Google Takeout service. Even after the organization detected the intrusion and successfully ousted the attackers from their network, Cold Relic’s campaign did not stop. Google observed the threat actor continuing to exploit stolen identities by deploying compromised email accounts to impersonate former employees of the victim organization. These accounts were used to target both internal personnel and external associates within the exact same professional circles, kicking off a renewed social engineering campaign designed to harvest additional intelligence.

Russian COLD RELIC Hackers Use Stolen Google Workspace Data to Breach Second U.S. Organization

The operation then transitioned into its next phase, shifting focus toward a second U.S.-based organization. This targeted entity utilized Microsoft Entra ID as its federated single sign-on identity provider for Google Workspace. To breach this second perimeter, Cold Relic leaned heavily on the intelligence gathered from their previous intrusion. By impersonating a recognized employee from the first compromised organization, the threat actor initiated contact with a targeted employee at the second entity using a secure messaging application.

Under the realistic pretext of discussing a technical issue or sharing a critical document, the attacker transmitted sophisticated phishing links designed specifically to harvest credentials and authentication tokens for Microsoft Entra ID. The familiarity of the purported sender drastically lowered the target’s guard, ensuring the success of the social engineering lure.

"Upon successful authentication using the stolen credentials, Cold Relic registered and used a new MFA device in the SSO solution to retain access to the user account throughout the intrusion," Google noted. By leveraging a persona already established within the victim’s professional network, the threat actor maximized their chances of a successful phishing outcome. Once inside the second organization, Cold Relic executed bulk downloads of all files accessible within the compromised user’s cloud drive.

Following the successful compromise of the second entity, the messaging application account utilized to initiate the initial conversation underwent a swift profile alteration, changing its display name and persona. Security researchers suspect this newly minted persona was immediately recycled to target additional organizations belonging to the same broader professional network, continuing the cycle of compromise.

Security analysts emphasize that these evolving campaigns illustrate a dangerous shift in the modern threat landscape, characterized by the seamless convergence of human-centric social engineering and sophisticated cloud identity abuse. By obtaining detailed strategic personnel rosters, corporate hierarchies, and operational documents from an initial victim, Russian intelligence services effectively convert legitimate business communications and trusted relationships into potent attack vectors against downstream partners. As organizations increasingly rely on interconnected cloud environments and federated single sign-on solutions, security experts warn that defending against multi-stage, identity-based pivot attacks will require heightened visibility across both administrative logs and external professional communication channels.