Malware Families Explained: Types, Behaviour and Response

A malware family is a group of samples built from the same code. How the main types differ, how analysts identify a family, and what to do.

A malware family is a group of malicious samples that share the same code, the same author, or the same tooling. Antivirus vendors give each family a name, and those names differ between vendors, which is why the same sample can be called three things by three products. What matters for defence is not the name but the behaviour the family reuses: if you know the family, you already know its persistence mechanism, its command-and-control pattern and often its detection opportunities.

What actually defines a family

Samples are grouped by shared code rather than by what they do to a victim. Two ransomware incidents can belong to different families while behaving identically at the end. Analysts group samples using file similarity (the same binary with small changes), shared strings and configs, the same C2 infrastructure, or the same builder kit sold on a forum.

That is why "a new ransomware" is rarely new. Most campaigns are an existing family with a fresh payload, which is good news: detections written for the family often survive the update.

The main families you will meet

Virus

Code that attaches itself to a legitimate file and runs when that file runs. Mostly historical in its classic form, because it needs a user to execute an infected file and it is easy to spot by file modification.

Worm

Self-propagating malware that spreads over a network without any user action, usually by exploiting a service. Worms are what turn one unpatched server into an entire estate in hours. Conficker and WannaCry's propagation component are the reference examples.

Trojan

Malware disguised as something the user wants — an installer, a crack, a document. A trojan is defined by its delivery, not its payload: the payload behind it is usually a backdoor or an infostealer.

Ransomware

Encrypts or exfiltrates data and demands payment. Modern ransomware is usually double or triple extortion: encryption for downtime, stolen data for a second demand, and sometimes a denial-of-service threat on top. The encryption step is now the last stage of an intrusion that may have started weeks earlier with a stolen VPN credential.

Rootkit

Hides itself and other malware by subverting the operating system — hooking system calls, hiding processes and files, or living in firmware. User-mode rootkits are detectable with standard tooling; kernel and firmware rootkits often require reimaging the machine rather than cleaning it.

Infostealer and spyware

Harvests credentials, session cookies, browser wallets and keystrokes, then sells or uses them. Infostealers are the most common malware in credential-stuffing chains: a stolen cookie can bypass the password and sometimes the second factor entirely.

Botnet

A network of infected machines controlled from a command-and-control server, used for spam, DDoS, proxy traffic or as a foothold for sale. Botnets are a business model rather than a payload.

Fileless and living-off-the-land

No malicious file on disk. The attacker abuses what is already installed — PowerShell, WMI, scheduled tasks, certutil, mshta — so traditional file scanning sees nothing unusual. Detection has to come from behaviour: what a trusted binary is being asked to do, and by whom.

How analysts identify a family

  • Hashes and similarity. A known sample hash matches exactly; fuzzy hashing and import-table comparison catch modified variants.
  • Sandbox behaviour. Running the sample and recording file, registry, network and process activity shows the family's habits even when the binary is new.
  • YARA rules. Pattern-matching rules written against strings and structures that are stable across the family.
  • Infrastructure. The same C2 domain, the same TLS certificate, the same hosting range across samples links a campaign together.

If you think a machine is infected

  1. Isolate it from the network first. Containment beats analysis in the first ten minutes.
  2. Capture evidence before cleaning — memory, disk image, and the logs that will roll over.
  3. Assume every credential used on that host is exposed, and rotate them from a clean machine.
  4. Do not pay. Payment funds the next campaign and does not guarantee a working decryptor.
  5. Rebuild rather than clean when a rootkit or kernel-level component is involved.

Where the family name stops being useful

A family name tells you how the malware works, not what happened to you. Two organisations hit by the same ransomware family can have completely different root causes — one a phishing click, the other an exposed RDP port. Track the family for detection, but investigate the intrusion for the root cause.

Related reading: EDR and XDR for the telemetry that catches these families, lateral movement for what happens after the first host falls, and the first-hour incident response cheat sheet for the order of operations.


Keep reading

Related articles

Browse all articles