Russian FSB-Linked Gang Uses One Cloud Breach to Hit a Second U.S. Org

A Russian state-sponsored hacker group has orchestrated a campaign where an initial cloud breach of one U.S.-based organization served as the springboard for a second attack in 2026. The breach involved Google Workspace and exposed employee directories and sensitive files, allowing the attackers to pose as employees and infiltrate a second U.S. organization using stolen credentials.

One breach feeding into the next

The group, identified as COLD RELIC (also known by names like COLDRIVER, Star Blizzard, and Callisto), compromised a U.S. organization’s Google Workspace account in late 2025. Using elevated access granted by a Super Admin account, the attackers stole an entire cloud drive, exported the staff directory, took tenant-wide exports via Google’s Customer Takeout feature, and deleted security alerts to cover their tracks. Some passkeys they enrolled survived password resets because no one had revoked them manually.

After securing those resources, the threat actors impersonated employees and leveraged their connections to target a second organization in 2026. In this follow-up attack, the group exploited Microsoft Entra ID—used by the second target for single sign-on into Google Workspace—and used social engineering conversations initiated over an encrypted messaging app rather than regular corporate email. MFA fatigue attacks, stolen credentials, and manipulated trust relationships formed the core of their approach.

What tools and tactics were used

COLD RELIC’s breach path included a progression of methods: repeated MFA prompts until approval (known as MFA fatigue), use of residential proxies to make login activity appear legitimate, enrolling unfamiliar devices and passkeys, and exploiting single sign-on configurations. They also targeted phishing redirects tied back to the first victim. Google discovered that the group’s previous campaign in 2025 had involved malware known as LOSTKEYS, though no specific malware payload was reported in these two breaches. Instead, credential theft and impersonation proved just as effective.

Investigators have published a list of indicators of compromise (IoCs) to help other organizations detect such activity. These include suspicious MFA behavior, device registrations from unfamiliar networks, bulk downloads or external sharing of files, customer takeout exports, staff directory exports, unexpected use of backup codes, alert tampering, and searches for credential files among cloud storage.

COLD RELIC has been officially linked by U.S. authorities to Russia’s Federal Security Service (FSB). Google’s researchers observed ongoing phishing activity from this group as recently as November 2026, confirming that the operation is still active.

This attack underscores a new shift: cloud breaches are no longer isolated incidents. In this case, first victim’s records became a weapon to breach another target—all without needing classic malware in the second phase.

Why it matters: for organizations relying on cloud infrastructure and single sign-on tools, this is a wake-up call. Tactics like MFA fatigue, passkey persistence, and credential-based impersonation are now central to high-end threat actor toolkits. If your org hasn’t yet audited its trust relationships, device enrollment policies, and response protocols for passkeys or backup codes, those gaps may already be under threat. Monitoring for the listed IoCs and enforcing stricter access alongside frequent audits isn’t optional—it’s essential going forward.