Some Ransomware Crews Wipe the Backups Too · Would Yours Survive the Same Attack?

A new CISA and FBI advisory says one ransomware group deletes network backups before it encrypts anything. If you are just hearing about this, the tests are worth running now: who can delete your backups, whether a real restore actually works, and who answers the phone during the next outage.
Ransomed After Backup Deletions

Picture a $60 million regional distributor on an ordinary Tuesday. The IT manager is about to take a few days off, the outsourced help desk runs light on weekends, and the owner asks one question before everyone leaves for the day: if our servers were locked on Friday night, could we be shipping again by Monday? The IT manager says yes, because the backups run every night. Both of them are telling the truth about the backups. Neither has checked whether the backups would still be there.

Both of them are telling the truth about the backups. Neither has checked whether the backups would still be there.

That second question matters more this month than it did a year ago. On October 20, 2026, CISA and the FBI, with the Secret Service and partner agencies abroad, published a joint advisory on a ransomware group called SafePay. Most of it is a technical catalog for defenders. One passage is written, in effect, for owners, because it describes attackers who go after the recovery copies before they lock anything.

 

What the Advisory Actually Says

The advisory says SafePay has been highly active since August 2024 and targets businesses and critical infrastructure across North America, South America, Europe and Australia. For a way in, the group relies on stolen or guessed passwords, flaws in security devices, and phishing, and the investigators saw it use exposed credentials, brute-force password attacks and VPN access. Once inside, it creates or takes over accounts, uses ordinary administrative tools to move around, and switches off security protections.

Then comes the part that concerns an owner most. According to the advisory, the group halts the Volume Shadow Copy Service, a Windows feature built to help recover a corrupted system, and resets network-attached storage that holds backup data to its factory settings. In some cases, it says, the actors deleted all network-attached backups of company data. The advisory describes encrypting files and leaving a ransom note that demands payment in cryptocurrency as the end of that sequence, after data has been copied out of the network.

Two cautions on how far to take this. First, the advisory describes one group, and we do not know how typical this backup behavior is across ransomware crews in general. Second, its stated audience is government and critical infrastructure defenders, although it says plainly that the group targets businesses. Our judgment is that the specific group matters less than the pattern. An attacker holding an administrator’s credentials can reach whatever that administrator can reach, and in our experience that often includes the backup copies.

 

Why Holidays and Weekends Are Part of the Picture

A 2021 CISA and FBI release made a related point. It said malicious actors have often taken advantage of holidays and weekends to disrupt networks, and it asked organizations to identify IT security staff who would be available on weekends and holidays if an attack came. That release is now archived, and it said itself that it rested on the timing of earlier attacks rather than on a specific threat report. We cite it as a pattern, not as a forecast about this year.

The practical reading is plain. Any long weekend or holiday is a stretch when more people are away and fewer are watching. If your recovery depends on one person who is traveling, or on a provider whose holiday staffing nobody has read, you have a gap that a calendar can predict even when an attacker’s schedule cannot be predicted. If this is the first time you have thought about it, the moment to check is now, not after the next alert lands.

 

Picture the Backup Chain

Go back to the distributor. Its backups are real. A backup server copies the file server and the accounting system every night, and a network storage box in the same closet keeps two weeks of copies. A second copy goes to a cloud storage account once a week. All of this works, and an auditor would find the nightly job green every day for a year.

Now ask who can delete each copy. The backup server is joined to the same Windows domain as everything else, so a domain administrator can reach it. The storage box answers to a login the IT manager set up years ago and shares with the managed service provider. The cloud account uses the same sign-in as the rest of the company’s administrators, with no second approval needed to delete anything. In this picture, one stolen administrator credential reaches all three. The backups are plentiful, and they are not independent of each other.

That is the question we would ask before any other. Not whether you have backups, but whether the account an attacker is most likely to steal can delete them.

Not whether you have backups, but whether the account an attacker is most likely to steal can delete them.

The advisory’s own recommendations point the same way. It asks for offline backups, and for backup data that is encrypted, immutable (in its words, “cannot be altered or deleted”), and covers all of the company’s data. It also asks for multiple copies kept in a physically separate, segmented and secure location.

 

Three Tests You Can Run This Week

None of these needs a new product. The first is a reach test. Ask your IT lead, or your provider, to write down every place a copy of your critical data lives and, for each one, which accounts can delete or overwrite it. If the answer is the same short list of administrator accounts every time, you have the problem in the picture above. A good answer names at least one copy that no everyday administrator login can change, whether that is an immutable storage setting at a provider, a vault with its own separate sign-in, or a drive that is physically disconnected.

The second is a restore test, and the standard is a real restore rather than a green status light. Pick one important server or shared folder, restore it to a spare location, and time it. Write down the minutes, the person who did it, and anything that surprised them. In our experience the first restore test turns up something: a database nobody had included, a stale credential, a recovery step only one person knows. Better to find that on a Wednesday than on a Saturday. We asked a shorter version of this question in our piece on questions to ask before you approve next year’s IT budget, and the advisory itself says to test backup and restoration processes regularly.

The third is a weekend test. Ask who is on call for the next long weekend or holiday, by name and mobile number, and ask what that person is allowed to do without calling you: isolate a server, disable an account, phone the cyber insurer’s hotline. If your IT is outsourced, ask the provider for its holiday coverage in writing. If the honest answer is that someone will see the alert on Monday, you have learned what a quiet Friday night would look like.

 

What a Good Answer Sounds Like

Vague answers are the warning sign. “We back up nightly” describes a schedule, not a result. A good answer has the shape of a date, a duration and a name: we restored the accounting database on a stated date in two hours and ten minutes, the oldest protected copy is 30 days old and nobody on staff can delete it, and the person on call over the holiday is named and answered a test call last month. Those figures are illustrative, but the shape is what you are listening for. If you hear the shape, you can stop asking. If you hear reassurance, ask again in different words.

If you hear the shape, you can stop asking. If you hear reassurance, ask again in different words.

It also helps to decide, before anything happens, what recovery is worth to you. How many days of orders can the business absorb without shipping? Which two systems have to come back first? An owner who has answered those questions in November can make a calm decision in the first hour of an incident instead of an expensive one in the sixth hour. Our cybersecurity work starts with those priorities for the same reason.

 

What This Will Not Do

A tested backup does not keep attackers out. The advisory’s own key actions are about the front door: patch firewalls and other edge devices, restrict administrative tools and privileges, and use multifactor authentication. Who owns those devices is often the real gap, and we wrote about that ownership problem for field networks in a piece on who owns the modem on a well pad. The same question applies to the firewall in your office.

It also does not answer the theft problem. The advisory says the group copies data out of the network before it encrypts, so a perfect restore can still leave you with a disclosure to handle, and with decisions about customers, regulators and your insurer that a backup cannot make for you. And the advisory covers one group; others behave differently. If your servers are old enough to have lost security updates, everything above gets worse, and our note on Windows Server 2016 and the January 12 deadline covers that sorting job. This is not legal advice, and it does not replace what your insurance policy requires; ask your broker what the policy says about backups.

 

Where to Start

Take one step this week. Ask your IT lead to restore one real shared folder from your most recent backup, time it, and give you the result in writing, along with a list of the accounts that could have deleted that backup and the name and mobile number of the person on call over the next holiday weekend. If the restore fails, or nobody can say who is on call, you have your first agenda item for a Monday meeting, and a cheap one. If this is the first time you are hearing that backups can be deleted this way, treat that as reason to move faster, not slower: isolate at least one copy of your backups and confirm it is protected before you do anything else.

If you want an outside view of how recoverable you really are, an assessment is one way to get it, and CIO IQ gives you an experienced CIO’s advice at $1k per week. You can also reach us through the contact form on our about page.

More from our blog

Your AI Dividend Can Fund a Reputation · A Role-Model Layoff

Your AI Dividend Can Fund a Reputation · A Role-Model Layoff

An AI-driven layoff can be announced like any other, or funded differently, with the savings paying for a real bridge…
U.S. Cyber Insurance Stopped Getting Cheaper · What's Negotiable?

U.S. Cyber Insurance Stopped Getting Cheaper · What's Negotiable?

U.S. cyber insurance rates finally stopped falling, but claims keep rising. We walk through what insurers now treat as standard,…
Who Owns the Modem on Your Well Pad? · The Cyber Gap Between IT and Field Operations

Who Owns the Modem on Your Well Pad? · The Cyber Gap Between IT and Field Operations

Many mid-market oil and gas operators split technology between IT in the office and operations in the field, and the…