GUIDE · RANSOMWARE RECOVERY

How to check that backups are not already compromised before restoring

Before you bring data back into use, you need to know which copy is clean. This guide explains why a backup may already be compromised, how to check it with concrete tests and how to document the choice. By the end you will know how to set up the verification, recognise the results that require further investigation and decide which copy to restore from.

Updated october 2026
8 min read
Altovar Technical Editorial Team

1. Why a backup can already be compromised

Many ransomware attacks do not stop at encrypting data at the moment they are discovered. Often the attacker stays in the network for an extended period, observes, collects credentials and looks for backup systems. If they manage to reach them, they can alter or encrypt the copies too, or plant code that only activates after a certain time. For this reason, the latest available copy is not automatically the right one.

The problem is that the date of the infection often does not coincide with the date of the visible attack. To choose a copy, you therefore need to reconstruct, with the help of those analysing the incident, when the intrusion began, and then verify that the copies made before that moment are intact. This is not something to improvise during the crisis, and it deserves dedicated time and skills.

2. Prerequisites: what you need before starting

The verification requires three things: an isolated environment in which to analyse the copies, access to the logs of the backup system and a timeline of the incident, even a provisional one. The isolated environment is needed because the copy must not be mounted on the network that could still be compromised. Without these elements, the verification is reduced to a check on the date, which alone is not enough to say whether a copy is clean.

You also need a list of the available restore points, with the date of each and the type of copy. If the backup system does not provide this list clearly, that is already a first problem to report to the supplier. Also ask whether offline or immutable copies exist, that is, copies that cannot be modified with the credentials used day to day, because they are the most useful in this scenario.

3. The steps of the verification

The verification must be carried out in sequence, starting from the copies closest to the likely infection and working back in time. Each copy that passes all the checks becomes a candidate for restore. Those that fail must be isolated and documented, because the failure is information about the attack, not just a technical obstacle to get around. The basic criterion is simple: each step must leave a result that someone else can review later.

The work can be carried out by the IT supplier or by an incident response team, but the business owner must understand the sequence and approve the final choice. Every check must be recorded with its date, the person who performed it and the result, because it will be needed for the incident documentation and for later reviews. Do not look for shortcuts: a copy chosen without minimum checks can bring the problem back into the clean network.

  1. Identify the presumed date of the intrusion based on the logs and analyses available.
  2. List the restore points prior to that date, from the most recent to the oldest.
  3. Mount each candidate copy in the isolated environment, read-only where possible.
  4. Run a scan with up-to-date tools and record every finding.
  5. Compare file integrity with the check values recorded at the time of the backup, where available.
  6. Verify that the test system starts and that the data is readable without errors.
  7. Choose the most recent copy that passes all checks, and record the reasoning.

4. Integrity and content checks

Integrity checks help establish whether a file has changed since it was copied. When the backup system records check values, the comparison is direct. When it does not, you can use values calculated on a reference copy of known date, if one exists. A file that is encrypted or visibly altered will fail the check, but a subtle alteration may slip through, which is why content scanning is complementary.

Content scanning looks for suspicious code, unexpected scripts, executable files in anomalous locations and modified configurations. For this reason the verification should be combined with log analysis and the judgement of those who know the systems. A negative result is an indication to investigate further, not a final verdict. Even an ambiguous result should be recorded with its uncertainty, because it helps the decision-maker choose with greater awareness.

CheckWhat it revealsLimitation
Copy dateWhether the copy predates the estimated date of the intrusionDepends on the quality of the estimate
Check valuesWhether files have changed since the backupRequires values recorded at source
Content scanPresence of suspicious code or scriptsDoes not guarantee the absence of everything
Start in test environmentWhether the system works without anomaliesDoes not verify the data in depth

5. Common mistakes in verification

The first mistake is trusting the date shown by the backup software without checking that it matches the actual content. The second is mounting the copy directly on the production network to save time, risking reinfecting the systems. The third is skipping the verification because the most recent copy looks fine, precisely where the attacker is most likely to have acted.

A less obvious mistake is considering the verification complete at the first positive result. A check may find nothing because of the limits of the tool, and a copy may pass the scan yet contain a configuration change. Document which checks you have carried out and which you have not, so that whoever decides knows exactly what the choice rests on. A frequent error is also trying to fix the start-up of a test system without first understanding why it stopped.

6. How to know the verification has worked

The verification has succeeded when you have a chosen copy, a written rationale and a list of the checks performed with their respective results. It is not enough for the restore to start: the choice must be explainable to those who will ask you about the incident. If the verification has produced anomalies, the useful question is which part of the process generated them, not just whether the copy is acceptable.

Before bringing data back onto the main network, review the results with the manager and with whoever analysed the attack. If doubts emerge, it is better to move the choice to an older copy than to risk bringing the infection back in. The delay of a verification is a known cost, whereas a reinfection is a cost that repeats. Always present the result with its limits, because no check covers everything, and it is better to state this upfront.

7. How to start: the first check this week

This week, ask the supplier for the log of available restore points and check whether there is at least one offline or immutable copy. If the answer is not clear, that is the first problem to solve, because without a reliable list the verification becomes a blind search. Even a simple table with dates and types of copy is already a step forward.

Once a candidate copy has been identified, prepare the isolated environment and agree the sequence of checks with the manager. Do not start mounting copies on the production network to gain time: the time gained that way is lost with the first reinfection. After the first step, arrange a short meeting with the managers to decide who carries out the verification and with which tools.

Frequently Asked Questions

The backup software reports no errors: is the copy clean?

No. A backup without errors shows that the copy was written, not that it is intact or free of malware. A file may have been altered before it was copied, or the copy may contain code that activates later. For this reason a specific verification is needed, which does not depend on the outcome reported by the backup software.

How do I estimate the date of the intrusion if the logs are incomplete?

You work through successive approximations, cross-referencing the available logs, alert messages, account creation dates and configuration changes. When logs are missing, the estimate should be made cautiously and with the input of those analysing the incident. The result is a working hypothesis, to be revised as new evidence emerges.

Can I use the oldest available copy to be safe?

It can be a prudent choice, but it has a cost: the loss of data produced between that copy and the incident. It must be assessed case by case with the process manager, because the loss of recent work may be more acceptable than a reinfection. The choice should be justified in writing.

Who should carry out the verification?

A technician with forensic analysis or incident response skills, supported by the IT manager who knows the systems. The business owner or a delegate approves the final choice. Separating execution from approval makes the process more robust, because the decision does not depend on a single point of view.

What do I do if all the copies are compromised?

This is a serious scenario that requires a different approach: deeper analysis, possible reconstruction from alternative sources such as external archives, original documents or third-party systems, and an assessment of notification obligations. It must be handled with an incident response team and a legal adviser, without improvising solutions.

About this content

General informational guide. Regulations and technical requirements change over time: always check the official sources and assess your specific case with a qualified technical or legal advisor.

GET IN TOUCH

Talk to our team.

We'll show how Altovar integrates into your infrastructure. No commitment, no sales pitch. Just a technical conversation.

BOOK A DEMO →