Catching Ransomware: Introducing encrypts-files and Stricter MITRE ATT&CK T1486

When we search for ransomware samples during an investigation or detection test, we usually want to answer a direct question: Did this file actually encrypt user documents when executed?

Modern ransomware families implement file encryption in many different ways. While commodity builders may call standard operating system cryptographic APIs, contemporary Ransomware-as-a-Service (RaaS) payloads written in Rust, Go, C++, or .NET frequently embed statically linked AES, ChaCha20-Poly1305, RSA, or custom intermittent encryption routines that never touch OS crypto libraries. At the same time, benign installers, browsers, and updaters legitimately use cryptography and write thousands of temporary files during normal execution.

This week, we rolled out three updates across VirusTotal and Google Threat Intelligence to identify malware that encrypts user files:

  1. New encrypts-files Tag Across Windows and macOS Sandboxes: A cipher-agnostic behavioral trait (behaviour_tags:encrypts_files) and file-level search tag (tag:encrypts-files) emitted across CAPE Sandbox, Zenbox, and macOS Mirage only when dynamic analysis confirms real encryption of user documents.
  2. Stricter MITRE ATT&CK T1486 Gating in CAPE Sandbox: Previously, generic CAPE file-churn signatures mapped high-volume file deletes, overwrites, or moves directly to T1486. Newly scanned CAPE reports now require verified user-file encryption before emitting those T1486 mappings, cutting CAPE-only daily T1486 hits from 1,039/day (78.5% with 0 AV detections) to 0 on benign files and reducing false positives across platform-wide T1486 volume by 97.6% (from 7,238 to 171 files per day).
  3. Real-Time VirusTotal Livehunt Support: Real-time YARA-X matching on sandbox reports via vt.BehaviourTrait.ENCRYPTS_FILES.

Figure 1: High-level overview of how CAPE Sandbox, Zenbox, and macOS Mirage verify user-document encryption before emitting tag:encrypts-files and gating CAPE T1486.

How Cipher-Agnostic File-Encryption Verification Works

Across CAPE Sandbox (Windows), Zenbox (Windows), and macOS Mirage (Apple Silicon), each sandbox environment contains realistic user documents across standard user profile directories.

During and after detonation, our sandboxes apply three verification gates before emitting encrypts-files (and before CAPE emits generic T1486 mappings):

  1. Content Destruction and Entropy Verification (Cipher-Agnostic): Rather than hooking OS cryptographic APIs, each sandbox inspects modified user-file content on disk using format-aware verification tailored to its platform. Depending on the file type and sandbox, checks evaluate destroyed file structures, elevated entropy, or non-text byte patterns alongside corroborating ransomware behaviors without relying on a single fixed entropy threshold. Because verification evaluates the resulting files on disk, it catches AES, ChaCha20, intermittent encryption, and XOR ciphers alike.
  2. Coordinated Ransomware Extension Renames: Many ransomware families read original user documents, write encrypted copies under a uniform new extension (for example, .k404, .lkd, .INC, .BLWZ, or .heartbreaker), and delete the originals. Our sandboxes correlate reads, writes, moves, and unlinks across user directories to confirm coordinated encryption renames across user documents.
  3. Benign File-Churn Exclusion Guards: Self-extracting archives, software installers, compilers, browser caches, and logs are explicitly excluded when they only generate routine temporary or application files rather than encrypting user documents. Similarly, dropping a README.txt file or calling crypto APIs alone never triggers encrypts-files without corroborated document encryption.

Real-World Ransomware Caught in Action

Across 328,257 newly submitted unique files (405,469 total files including rescans) analyzed in CAPE Sandbox, Zenbox, and macOS Mirage since November 17, tag:encrypts-files matched 55 files and surfaced in-the-wild and validation payloads across Windows executables, Python scripts, DLLs, compressed archives (ZIP and RAR), and a rescanned Apple Silicon macOS binary with zero benign software false positives across all 55 manually reviewed matches (alongside 0 benign false positives across 9,976 random CAPE reports and 400 macOS Mirage runs in pre-rollout backtesting). Below are three examples you can inspect on VirusTotal.

1. Catching Zero- and Low-Detection In-the-Wild Lockers: The .k404 Campaign and lkd.exe

Static signatures often lag behind script-based or freshly compiled lockers. In an active .k404 extortion campaign submitted on November 17-18, Python locker fb0982bb9f5d5700685dcf4e99ff695f4593fbc3e3d0359d4c2baf23c545f354 (system_update.pyw) was flagged by 0 out of 75 static antivirus engines at the time of analysis, while companion PE binary 3dd45ab8e2f9c1af773a75221d792fb78870764419170e9c9c5941160364c700 (K404T_Fidye.exe) had 10 / 75 detections and separate Windows locker a8c2402ef3ae115300f5428712ccb0b5461f4a2f14b5bfdfbe08d5

[…]
Content was trimmed to protect the source. Please visit the original article for the full text.

This article has been indexed from VirusTotal Blog