The COLD RELIC Google Workspace breach began with stolen Super Admin credentials and an MFA push fatigue attack. A Russian espionage group took over an entire US organization’s Google Workspace tenant in late 2025, downloaded its whole cloud drive, and then used the staff directory and relationships inside it to impersonate an employee and breach a second US organization in 2026. Google’s Mandiant and threat intelligence teams published the case study on 17 November, and the dated chain it describes runs from late 2025 to November 2026, about a year by TechReview’s count.
The group is COLD RELIC, formerly tracked as COLDRIVER and UNC4057 and also known as Star Blizzard, Callisto and Gossamer Bear. The US Department of Justice has assessed that it operates within Russia’s Federal Security Service, the FSB, according to the Google Cloud Blog post by Mandiant and the Google Threat Intelligence Group. The group targets government officials, defense contractors and non-governmental organizations across Europe and North America, seeking information of strategic or political relevance to the Russian government, and it uses compromised access to reach its victims’ wider social and professional networks, the blog says. Google’s team has tracked and disrupted the group’s phishing campaigns since 2015.
How the COLD RELIC Google Workspace breach unfolded
The operation had two parts, and the second only worked because of the first. In late 2025, COLD RELIC compromised a US organization’s Google Workspace using stolen credentials for a Super Admin account and a multi-factor authentication push fatigue attack. With that access, the group downloaded the organization’s entire cloud drive, exported full employee lists and ran a domain-wide data export. In 2026, it impersonated an employee of that first victim on an encrypted messaging app and used the pretext to phish a second US organization, stealing Microsoft Entra ID credentials and tokens and bulk-downloading the files one user could reach.
Both victim organizations are unnamed, and no affected-user count has been disclosed. The Hacker News covered the same case study and corroborates the account, including the Super Admin access, the full-drive download, the employee lists, the Customer Takeout export and the Entra ID theft.
The first breach: a Super Admin account and an accepted MFA request
Initial access came through a residential proxy network in the United States located close to the victim, plus the stolen credentials of a Super Admin account combined with an MFA push notification fatigue attack, according to the Google blog. The victim accepted an MFA request, after which the attackers registered a device model previously unseen in the tenant.
The attackers kept access with passkeys, MFA backup codes and rogue MFA devices. Using the Super Admin account, they downloaded the contents of the entire organization’s cloud storage drive, by mass-sharing files to a COLD RELIC-operated email account and by downloading files directly. They also exported full employee lists and started a domain-wide data export through Google Workspace Customer Takeout using Super Admin privileges.
While inside the tenant, the group accessed the Admin Console, queried Admin security investigation and email audit logs to check defensive visibility, and viewed, modified and deleted Alert Center notifications to remove forensic evidence of the initial intrusion.
COLD RELIC held access to the first victim for several months. It synced email inboxes via OAuth to external email clients, re-authenticated with attacker-enrolled passkeys that persisted across password and session resets and bypassed device registration controls, and ran targeted cloud-drive searches for credentials, email archives, personnel information and other intelligence-collection material.
After being removed from the first victim, the group used email accounts impersonating that organization’s employees to target internal employees and associates in the same professional circles with renewed social engineering attempts, the blog says.
The second breach: a familiar name on a messaging app
In 2026, Mandiant investigated a second intrusion at a separate US organization that used Microsoft Entra ID as its federated single sign-on provider for Google Workspace. COLD RELIC impersonated an employee of the first victim to establish a pretext for communications over a secure messaging app, then shared phishing links designed to steal credentials and authentication tokens for Entra ID, redirecting the victim to document lures about a technical issue.
Once the victim authenticated with the stolen credentials, the group registered and used a new MFA device in the single sign-on solution to keep access to the user account, and bulk-downloaded files accessible in the compromised user’s cloud drive.
Google’s team correlated a COLD RELIC redirector URL from the second intrusion that had been submitted to VirusTotal in June 2026, whose query parameter referenced the first victim organization. The June 2026 date is, in our reading, evidence the follow-on operation was already running by mid-2026.
After the second compromise, the encrypted messaging application used for the approach recorded that the threat actor changed their profile name to a completely new person. Google’s team believes this new persona was likely used to continue targeting other organizations in the same professional circles. That is an assessment, not a confirmed third victim, and no third victim has been named.
Mandiant says it is likely that COLD RELIC used stolen contact and relationship information from the first organization to target the second, inferred from phishing URL parameters and impersonation pretexts. The blog frames that as an assessment, not proof.
One stolen tenant, about a year of payoff
The dated chain below is TechReview’s own reconstruction from the Google blog and the Justice Department record. It shows how long one compromise kept paying off.
- 2015: Google’s threat intelligence team begins tracking and disrupting COLD RELIC phishing campaigns. By 2026, that is 11 years of tracking.
- October 2024: the US Department of Justice seizes 41 internet domains used by the Callisto Group, an operational unit within FSB Center 18, while Microsoft restrains 66 more in a concurrent civil action. That is 107 domains in one coordinated takedown, our calculation from the two figures.
- Late 2025: the first US organization is breached. Stolen Super Admin credentials plus MFA push fatigue open the tenant; the entire cloud drive is downloaded, full employee lists are exported, and a domain-wide Customer Takeout export runs. The attackers stay in for several months.
- June 2026: a COLD RELIC redirector URL referencing the first victim organization is submitted to VirusTotal, dated evidence that the follow-on operation was already running by mid-2026.
- 2026: the second US organization is breached through an impersonated employee on an encrypted messaging app, Entra ID credential theft and a newly registered MFA device.
- November 2026: Google’s team is still observing new COLD RELIC phishing, and the case study is published on 17 November.
The arc is the point. A single late-2025 compromise supplied the staff directory, relationships and pretexts that powered a second breach months later, and the renamed messaging persona suggests the chain may not have stopped at two. That last part is Google’s assessment, not a confirmed fact.
Super Admin versus one user: how far each breach reached
The two intrusions show what administrative privilege is worth to an attacker, and what its absence costs them. The comparison below is TechReview’s own, drawn from the Google blog.
| First organization (Super Admin) | Second organization (one Entra ID user) | |
|---|---|---|
| Access level | Super Admin, tenant-wide | One compromised user account |
| Data taken | Entire cloud drive, full employee lists, domain-wide Customer Takeout export | Only the files that single user could access |
| Defensive moves | Audit logs queried, Alert Center notifications deleted | None reported |
| Persistence | Months; passkeys surviving password and session resets | New MFA device registered on the account |
Restricting administrative privileges limited the blast radius, but it did not stop the breach. One stolen employee identity still exposed that user’s files. Least privilege shrinks the damage; it is not a substitute for phishing-resistant MFA. That is our reading of the two cases, not Google’s wording.
Why a password reset did not end it
The case study carries a blunt lesson for anyone who has ever treated a password reset as the end of an incident. Attacker-enrolled passkeys persisted across password and session resets and bypassed device registration controls, according to the blog. In at least one case, a victim organization reset passwords for impacted accounts but failed to revoke all existing sessions, so COLD RELIC kept its access.
Google recommends revoking all active sessions and refresh tokens, removing unauthorized authenticators, and reviewing third-party application permissions across both Entra ID and Google Workspace, plus phishing-resistant MFA such as FIDO2 security keys. A password change on its own is explicitly insufficient, and the blog says so in those terms.
Credentials and other secrets stored in the victims’ cloud storage amplified both intrusions, the blog says, letting COLD RELIC pivot to additional accounts or services, including users’ personal accounts that employers typically do not monitor.
Google’s team has observed evidence of new COLD RELIC phishing operations as recently as November 2026, and Google issues Government-backed attack alerts when it detects COLD RELIC targeting Google accounts.
What to watch next
Whether either victim organization is named or speaks publicly. Both are unnamed as of 18 November, and no affected-user count has been disclosed. Whether a third linked victim surfaces, which the renamed messaging persona points toward but does not confirm. Whether mainstream tier-1 outlets pick the story up beyond cybersecurity trade press such as The Hacker News. Whether CISA or the FBI issue an advisory citing COLD RELIC or this Workspace and Entra ID tradecraft. And whether Google publishes indicators of compromise or names the messaging app used for the approach.
The COLD RELIC Google Workspace breach is a different campaign from the n0n extortion wave that hit Visa and Eli Lilly, which used third-party vendor data claims rather than cloud-identity espionage, and it is not part of the US tech outage wave, which was a set of service disruptions rather than an intrusion. The case study describes no Google Workspace vulnerability. The documented access path was stolen credentials, MFA push fatigue and cross-cloud identity abuse. The October 2024 takedown of 107 domains is background on the same group, not part of this incident, since the 2025-26 intrusions happened after it.


