Breaches and incidents

Russian group Star Blizzard used one hacked US firm to break into another

Google says the FSB-linked group, tracked as Cold Relic, stole identities and contacts from a US organisation breached in late 2025 and used them to phish a second one in 2026.

By Junglewise News DeskPublished 6 min read
Close-up of tangled black cables connected to a wall of modular server or display panels
Photo by Kier in Sight Archives on Unsplash

A Russian state-linked espionage group spent months inside one US organisation's cloud environment, took its files and employee lists, and then used the identities it found there to break into a second US organisation. Google Threat Intelligence Group (GTIG) and Mandiant, Google's incident response unit, published the analysis on 17 November 2026, according to Infosecurity Magazine.

The group is tracked by Google as Cold Relic and is better known as Star Blizzard, with Callisto, Coldriver, UNC4057 and Gossamer Bear in circulation as well. The US Department of Justice has assessed that the group operates within Russia's Federal Security Service (FSB), and Google says it has been highly effective against high-value targets including government officials, defence contractors and non-governmental organisations across Europe and North America. Neither victim organisation has been named, and The Record reported that Google did not say what information was taken.

The consequence for the second organisation is the part worth dwelling on: it was not breached through a software flaw or a weak password of its own. It was breached because someone it had reason to trust was, months earlier, and the attacker inherited that trust along with the contact list.

What to do now

  • Enforce phishing-resistant multi-factor authentication, ideally hardware security keys, which cannot be intercepted by a proxy page or approved by a tired user hitting "allow" on a push notification. Google listed this among its recommendations.
  • Restrict and audit passkey registrations, and harden the procedure for registering a new MFA device, so that an attacker who holds a valid session cannot quietly enrol one of their own.
  • If you suspect a compromise, revoke all user sessions as well as resetting passwords and MFA. Google warned that "password changes alone are insufficient" to evict this actor, and in at least one case a victim reset passwords while the attacker kept hijacked sessions that were never revoked.
  • Audit OAuth grants and third-party application access. The attackers continued reading mail through third-party applications after the obvious access routes were closed.
  • Apply data loss prevention policies to cloud storage to warn on or block credentials and other sensitive patterns being stored or shared externally. Employees storing passwords in company cloud drives gave the attackers a route into further accounts, including personal ones.
  • Enrich authentication logs with residential proxy and VPN reputation data, and use conditional access to block or flag sign-ins from known proxy networks. Logging in from a residential address near the target is how the group avoided impossible-travel alerts.
  • Keep sessions short and privileges minimal, so that a stolen credential buys less. Joseph Spero, principal incident response consultant at Google, recommended both, Dark Reading reported.
  • Tell staff to verify unexpected approaches through a separate channel they have used before, especially when the message arrives somewhere unusual such as a messaging app. Wes Shields, a security engineer at GTIG, gave that advice.

How the first intrusion unfolded

The first organisation was compromised in late 2025 through stolen credentials for a Super Admin account in its Google Workspace environment. The attackers signed in through a US residential proxy close to the victim's location, then sent repeated MFA push requests until someone approved one, which let them register a device to the tenant.

With administrator rights they downloaded the organisation's entire cloud storage drive, by mass-sharing files to an attacker-controlled email account and by downloading them directly, exported full employee lists, and kicked off a domain-wide export using Google Workspace Customer Takeout. They also went through the audit logs to work out what defenders could see, then opened, edited and erased warnings in the Alert Center so that the traces of their arrival went with them.

Persistence was built on identity rather than malware. The group enrolled its own passkeys, MFA backup codes and MFA devices, which survived password and session resets and sidestepped device registration controls, and it stayed for several months, syncing mailboxes to outside email clients and pulling selected documents from cloud drives.

Google's analysis notes that the group hunts specifically for material that makes the next phishing email credible: rosters, contact lists, resumes and event schedules. Researchers said that once it has a foothold, the group "uses stolen emails and data to establish authentic pretexts and impersonate contacts when reaching out to subsequent targets".

The handover to the second victim

Being removed from the first victim's systems did not end the campaign. GTIG saw Cold Relic sending phishing mail from accounts posing as that organisation's employees to colleagues and to people in the same professional circles.

In 2026, an operator presenting as a member of staff at the first victim struck up a conversation with someone at a second US organisation over an encrypted messaging app, and partway through shared links. The links were built to steal credentials and Microsoft Entra ID authentication tokens, and they forwarded the target to document lures that matched the discussion already under way, giving the whole exchange an unremarkable surface. After authentication succeeded, the group registered a new MFA device in the organisation's single sign-on system to hold onto the account, then used the Entra ID credentials and tokens to reach the victim's Google Workspace environment through federated single sign-on, Dark Reading reported. Large volumes of data were downloaded from the second victim's cloud storage.

"By impersonating someone known in the same professional circles as the new target, COLD RELIC increased their chance of phishing success," GTIG researchers wrote.

Technical details

Timeline: first US organisation compromised in late 2025; second compromised in 2026. One of the redirector URLs used in the second intrusion turned up on VirusTotal in June 2026, carrying a parameter that named the first victim, which is what tied the two cases together. Google published its analysis on 17 November 2026 and says it has seen new phishing activity from the group as recently as November 2026. It has tracked the actor since 2015, and the actor has been active for over a decade.

Tradecraft: stolen Super Admin credentials; US residential proxy infrastructure near the victim to defeat impossible-travel detection; MFA push fatigue; rogue MFA device, passkey and backup code enrolment; mailbox syncing to external clients; third-party application access for continued mail reading; administrative console access to assess visibility and remove alerts; Google Workspace Customer Takeout for domain-wide export; phishing pages reached through compromised websites that screen visitors before passing them on, often with a link parameter naming a previous victim or lure.

After the second intrusion the operator changed their messaging app profile to an entirely new persona, which GTIG believes was probably prepared for the next set of approaches in the same circles. SC Media noted that The Hacker News reported the research first.

Neither victim has been named, and Google did not disclose what information was stolen.

Share this article

Related stories

Scan your domain free: see what attackers can see from outside, in about 30 seconds.

This article as Markdown. Atom feed.