Threat researchTracked by EncryptLayer intelligence

Uncovering ArcticRAT: The Evolution from Targeted Crypto-Stealer to AI-Powered Android RAT

Published by
zimperium.com
Published
16 Nov 2026
Indicators
none submitted

The report

zLabs tracks ArcticRAT, an Android RAT that uses xAI’s Grok to mine stolen SMS for OTPs and balances, plus HVNC, fake lock screens, and ADB abuse.

zimperium.com

Read the full analysis: Uncovering ArcticRAT: The Evolution from Targeted Crypto-Stealer to AI-Powered Android RAT (zimperium.com).

What the report says

Zimperium’s zLabs researchers report on ArcticRAT, an Android malware family that began as spyware aimed at users of 11 cryptocurrency apps and has evolved into an AI-powered remote access trojan. The report says the latest variants give operators full device takeover through Hidden VNC and raw TCP C2, and that threat actors added LLM APIs to the exfiltration pipeline to parse stolen SMS and app data, first seen in samples from around mid-July 2026.

  • zLabs identified 78 unique malware samples, 42 distinct dropper samples, 36 unique payloads and 21 C2 IP addresses, and cataloged 129 distinct commands. Elsewhere the report says analysis revealed 17 distinct C2 IP addresses, two thirds of them in Hong Kong and 20% in Germany.
  • The infection chain starts at malicious sites reached via malvertising, smishing or third-party forums. Most droppers used a landing page impersonating Google Play, and some payloads were delivered from AWS buckets. Lures impersonated Amazon, file managers, TikTok and crypto apps.
  • The dropper requests VPN service privileges to null-route network traffic, which the report says blinds Google Play Protect telemetry during staging. It shows a fake installation interface in a WebView, decrypts the payload from its assets folder, installs it through the Android session installer and then ends the VPN.
  • The payload abuses Android Accessibility Services. After accessibility is granted, it connects to the C2 on mainport 9999 and sends a DEVINFO JSON containing battery level, screen state and Wi-Fi status, plus the user’s wallpaper, which the report says is used to evade sandbox analysis.
  • The C2 selects attack routines based on the victim profile, using dedicated ports: HVNC 8888, mic streaming 5000, screen cast 5055 and the accessibility remote control channel 5058. Screen frames are sent as JPEG binaries via MediaProjection and VirtualDisplay. Samples from v6.3 through v6.5.2 show continuing development.

Named in the report: ArcticRAT, zLabs, net.iceflow.arcticlink, net.frostlink.iceforge, HVNC, VlockPJO.

This digest was written from the report text itself by a language model and checked by script: every number and CVE id in it appears in the report. The report is the source; read it for the full detail.

Indicators

The report is tracked for context; no indicator from it met the bar for blocking (for example, it only names shared hosting).

How this is tracked

EncryptLayer intelligence, A.R.P. Syndicate’s open threat feed, reviews each indicator against the published report before adding it (#42). This page is generated from that record and the report itself; the indicators are never rewritten.