One stolen admin login, then a second victim through a fake colleague
Russian spies used a stolen admin login to pose as staff and phish new targets
Google says a Russian state-linked group, COLD RELIC, broke into a US organisation in late 2025 using a stolen Super Admin login, then copied its cloud files and staff list. In 2026 it posed as an employee of that organisation to breach a second US organisation, and Google says the group was still running new phishing in November. Its stated targets are government officials, defence contractors and non-governmental organisations in Europe and North America.
What to do this week
- Check unexpected requests through a second channel.If a colleague, friend or client asks you to open a file, sign in or approve something, phone or message them on a number you already have. Don't reply in the same thread.
- Don't sign in from a link in a message.Type the address yourself or use a bookmark. The links in this campaign were built to collect sign-in details, and they opened documents that matched the conversation.
- Pause on new contacts who claim to be a colleague.A new profile on a messaging app that quickly shares a document or link is a reason to check first.
- Decline sign-in prompts you didn't start.If a prompt arrives for a login you didn't make, decline it. Then change that password and end active sessions on the account.
- Ask whoever runs your work accounts what they are checking.Ask them to look for the signs listed above, to require approval for new MFA devices, and to end sessions after any suspected break-in.
What happened
A Russian state-linked hacking group used one stolen administrator login to copy a US organisation's cloud files, then used what it found to pose as that organisation's staff. Google's threat intelligence team says the same group broke into a second US organisation in 2026. If a colleague sends you a message about a shared file or a sign-in link, check it through a separate channel before you open anything.
- Severity
- High for staff at targeted organisations. Nothing to install.
- Who it targets
- Government officials, defence contractors and non-governmental organisations in Europe and North America
- Attribution
- The US Department of Justice has assessed that the group operates within Russia's Federal Security Service (FSB)
- First US breach
- Late 2025, using a stolen Super Admin login
- Second US breach
- 2026, through a person posing as an employee of the first organisation
Google's threat intelligence team (GTIG) published its report on the group[1] on 18 November 2026. Google previously tracked it as COLDRIVER and UNC4057, and it is also known as Star Blizzard, Callisto and Gossamer Bear. The US Department of Justice has assessed that it works within Russia's Federal Security Service (FSB).
The first attack hit a US organisation in late 2025. The attackers signed in with the stolen login of a Super Admin account, which is the highest level of control in a Google Workspace organisation. They connected through a US residential proxy near the victim, a service that makes a sign-in look like it comes from a home internet connection nearby. They then kept sending sign-in approval requests to the account holder until one was accepted.
With that control, the group downloaded the organisation's whole cloud drive and exported its staff list. It also started a domain-wide export through Google's Customer Takeout tool, which packages an organisation's data in one go. It then viewed, changed and deleted security alerts in the admin system, which Google says was meant to remove evidence of how the group got in.
The group stayed for several months. It added its own passkeys (sign-in keys stored on a device), backup codes and MFA devices to accounts. Those stayed active through password and session resets. Google also says that in at least one case, a victim reset passwords but did not end existing sign-in sessions, and the attackers kept access. The group linked inboxes to outside email apps so it could read mail there.
Even after the organisation removed the intruders, the group kept writing to people from email accounts that pretended to be employees of the victim. It aimed these messages at internal staff and at people in the same professional circles.
The second attack started in 2026. A person posing as an employee of the first organisation contacted a staff member at a second US organisation over an encrypted messaging app. Over time they built up a conversation, then sent links to documents that matched it. The links were built to collect sign-in details and login tokens for Microsoft Entra ID, the sign-in service the second organisation used alongside Google Workspace. A token is stored proof that someone has signed in, so a stolen one can open an account without the password. Once in, the group registered its own MFA device, downloaded large amounts of files from the person's cloud drive, and changed its messaging profile to a completely new name.
Google believes that new profile was used to approach other organisations in the same circles, and it has seen new phishing operations from the group as recently as November 2026. The report does not name the victims. It also does not list the web addresses or IP addresses behind the campaign.
The full story2 more sections
Who is affected
Most families are not the group's target. Its stated targets are government officials, defence contractors and non-governmental organisations in Europe and North America. If you work in one of those roles, or sit on the board of an organisation that does, this report is directly relevant to you.
Everyone else can still be reached in the same way. The group's method is to take an account that belongs to someone you know, then write to other people from it. Google also says that in some cases, credentials stored in a company's cloud storage let the group reach staff's personal accounts, which an employer does not normally watch.
That makes a message from a known name worth a second look. It is not proof that the person sent it. If the message asks you to open a file, sign in, approve a prompt or share a document, check it first.
What administrators should check
If you run a Google Workspace account for a business, school or charity, Google's report lists signs to look for in the admin logs. Look for a Customer Takeout export that nobody started, large downloads from the cloud drive, files shared to outside addresses (especially free email services), new passkeys or MFA devices on accounts, and batches of deleted alerts. Google also recommends reviewing which email apps have been given access to mail, and removing any that nobody approved.
Google says hardware keys cannot be approved through fatigue attacks. If you find signs of a break-in, end active sessions, remove unknown passkeys, MFA devices and app access, and then reset passwords. Google says a password change alone is not enough.
Timeline
-
HARD publishes its article, at High.
After the news
- Every data breach who was hacked, and how to check if you’re in one
- Data breach settlements the claims open now, by deadline
References3
- ^ Google Threat Intelligence Group (GTIG) and Mandiant (). “COLD RELIC: A Case Study in Impersonation and Compromise”. Primary report: the group's aliases, DOJ/FSB attribution, targets, both intrusions, persistence, Google's recommendations and the absence of published IOCs.
- Help Net Security (). “Russian spies turn one cloud breach into a stepping stone for the next”. Independent coverage of the GTIG report; quotes the Super Admin, Customer Takeout and Alert Center passages and links the GTIG report.
- The Record (). “Russian cyber-espionage group used stolen cloud data to hack into organizations, Google says”. Independent coverage confirming the impersonation framing and the 18 November date. Read as a summary only.
Situation Room articles are updated as events develop: updates are added to the timeline, and never edited away. Follow them by email or RSS.

