Most phishing write-ups describe one victim. A report published Nov. 17 by Google Threat Intelligence Group (GTIG) and Mandiant describes a chain. According to the companies, a Russian espionage group used what it stole from one US organization in late 2025 to break into a second US organization in 2026. The data from the first breach included a corporate staff directory and real identities, and the group used it to impersonate people the second victim already knew and trusted.
Google calls the group COLD RELIC. Defenders may know it by other names. Google's report says it was formerly tracked as COLDRIVER and UNC4057 and is also known as Star Blizzard, Callisto and Gossamer Bear. Google says the US Department of Justice has assessed that the group operates within Russia's Federal Security Service (FSB). The company says the group has compromised government officials, defense contractors and non-governmental organizations in Europe and North America, and that GTIG has tracked and disrupted its phishing campaigns since 2015.
The techniques in the report are not new. What the report shows is how they fit together: social engineering gets the group in, abuse of cloud identity settings keeps it there, and each victim's data supplies the bait for the next one.
How the chain worked
Mandiant says it responded to linked intrusions over the past year. In Google's telling, the late-2025 compromise of the first US entity produced more than mailbox contents. Google says the group downloaded a large volume of data, searching for names, email addresses and email archives, and exported full employee lists. That kind of material is what a convincing pretext is built from.
Infosecurity Magazine, reporting on the two intrusions, quotes Google as saying COLD RELIC shows "convergence of human-centric social engineering and cloud identity abuse." The group uses stolen emails and data to establish authentic pretexts and impersonate contacts. In the second case, Google says responders found the group made contact over an encrypted messaging app while impersonating an employee of the first victim, then shared phishing links under the pretext of a document about a technical issue. The approach would not have looked like generic phishing: it came from a name the target could place in their professional circle.
This is why awareness training alone does not stop this kind of operation. "Check whether the sender is someone you know" fails when the attacker has the target's colleague directory and the identities of real people.
Staying in: rogue MFA, passkeys and backup codes
The second half of the report covers persistence, and administrators should read it closely. The Hacker News reports that the group registered rogue multi-factor authentication (MFA) devices, passkeys and MFA backup codes to establish long-term persistence. In the first intrusion, Google says initial access combined stolen Super Admin credentials with an MFA push-notification fatigue attack, and the victim accepted an MFA request. In the second, after harvesting Microsoft Entra ID credentials and tokens, the group registered a new MFA device in the single sign-on (SSO) solution.
Each of these is a normal account feature:
- An MFA device is the phone or authenticator a user approves logins with. If an attacker adds their own, they can pass the second-factor check whenever they like.
- A passkey is a cryptographic credential stored on a device that replaces the password. Google reports the group re-authenticated using unrevoked attacker-enrolled passkeys that persisted across password and session resets.
- Backup codes are one-time codes meant for users who lose their phone. Anyone holding a set has a ready-made bypass.
The common thread is that a password reset, the usual first step after a suspected phishing compromise, may not evict an attacker who has enrolled their own authentication factors. The account looks secured while the intruder keeps a valid way back in.
With that foothold, according to The Hacker News, the group collected and exfiltrated email, chat and cloud storage data, then turned to the victim's social and professional networks. Google says the group kept access to the first victim for several months, and that after it was removed, email accounts impersonating the victim's employees were used to target people in the same professional circles. That is how the cycle repeats: today's victim provides the contact lists and pretexts for tomorrow's target. The Hacker News headline identifies the stolen data as Google Workspace data.
Why this matters beyond the two victims
Neither victim organization is named in the source reporting, and nothing in the sources indicates that a software vulnerability was exploited. This is an identity and trust problem, not a patching problem. That has two consequences.
First, there is no update that fixes it. The controls that matter are how accounts enroll and review authentication factors, and how staff verify unusual requests from known contacts.
Second, your exposure depends partly on who you work with. If a partner, prime contractor, subcontractor or NGO you correspond with is compromised, your staff can be targeted with messages that are hard to tell apart from the real thing. Google says the group has compromised defense contractors and government officials, so companies that work with those sectors should treat themselves as within the populations it describes.
What It Means for Defenders
Who is affected: According to Google, COLD RELIC's victims include government officials, defense contractors and NGOs in Europe and North America. The two intrusions in this report were at US organizations. Any organization whose staff correspond with those sectors should assume it could be used as a stepping stone, or targeted with pretexts built from a partner's stolen data.
Patches and fixed versions: None apply. The sources describe abuse of legitimate cloud account features after social engineering, not exploitation of a product flaw. Google's report includes its own detection and hardening guidance, and defenders should treat that report as the primary reference. Infosecurity Magazine summarizes it as including phishing-resistant MFA (particularly hardware keys), restricting and auditing passkey registrations, conditional access that accounts for residential proxy and VPN IP reputation, and cloud storage data loss prevention policies.
What to check, based on the techniques described:
- Audit enrolled authentication factors, not just passwords. For high-value accounts (executives, staff with government or defense contacts, administrators), review every registered MFA device and passkey and every set of issued backup codes. Look for enrollments that don't match a user's known devices, or that appeared around the time of a suspicious message.
- Alert on new factor enrollment. Because rogue MFA devices, passkeys and backup codes were central to persistence here, a new authentication factor on a sensitive account should be reviewed, especially soon after a login from an unfamiliar location or device.
- Make incident response go beyond a password reset. When an account is suspected compromised, remove any authentication factors that can't be verified, revoke existing backup codes and active sessions, and only then reset credentials. Otherwise an attacker who enrolled their own factor may still have access.
- Watch for bulk access to mail, chat and cloud storage. The group collected and exfiltrated data from all three, according to The Hacker News. Unusual volumes of downloads, exports or mailbox access from a single account are worth investigating.
- Verify unusual requests from known contacts through a separate channel. The pretexts in this campaign were built from real data stolen from a trusted organization, so a familiar sender name proves little. A callback to a number you already have on file, not one taken from the message, works better than inspecting the message.
- Tell partners when you've been hit. If your organization finds a compromise, the people in your directory and contact lists are the likely next targets. Telling them early may break the chain.
The lesson of the report is that an organization's cloud identity is more than a login. It holds the relationships an attacker can use to reach the next victim. Protecting the authentication factors attached to those accounts matters as much as protecting the password.
Sources
- Google Cloud Blog (GTIG/Mandiant): COLD RELIC: A Case Study in Impersonation and Compromise
- The Hacker News: Russian COLD RELIC Hackers Use Stolen Google Workspace Data to Breach Second U.S. Organization
- Infosecurity Magazine: Russian Espionage Actor Star Blizzard Focuses on Cloud Identities for Mass Data Theft