The ToxicPanda Never Sleeps: ToxicPanda 2.0 Prepares its Next Strike on Mobile

zLabs identified ToxicPanda 2.0, an updated Android banking Trojan that greatly expands its command set, targeting scope, and fraud capabilities, including PIN theft, lock-screen credential harvesting, and abuse of Android Wireless Debugging for shell access. The campaign also shifts delivery to Amazon AWS-hosted buckets and targets 349 financial apps and 140+ banking and cryptocurrency apps across 16 countries. #ToxicPanda #AmazonAWS #AndroidAccessibilityService #ADB #SPAKE2

Keypoints

  • ToxicPanda 2.0 is an updated Android banking Trojan with 167 remote commands and significantly expanded capabilities.
  • The malware now targets more than 140 banking and cryptocurrency apps for PIN theft and 349 financial institutions for overlay-based credential theft.
  • It abuses Android Accessibility Service to capture UI elements, automate actions, and support fraudulent activity.
  • The malware can abuse Android Wireless Debugging (ADB) to obtain shell-level access and escalate privileges on infected devices.
  • It can steal lock-screen credentials by overlaying spoofed screens that mimic the native Android lock screen.
  • Distribution has shifted to Amazon AWS-hosted buckets, indicating use of cloud infrastructure for malware delivery.
  • Several previously unimplemented commands are now functional, including OEM permission automation, device administrator exploitation, transparent overlays, and auto-start bypass logic.

MITRE Techniques

  • [T1113 ] Screen Capture – The malware abuses Accessibility Service to steal UI content and screen elements from targeted apps (‘steal every UI element on the screen’).
  • [T1518.001 ] Security Software Discovery – It enumerates installed applications and sends package names/icons to identify banking and financial apps (‘enumerates all applications installed on the victim’s device’).
  • [T1056.001 ] Keylogging – The malware captures touches, PINs, patterns, and passwords entered by victims through overlays and touch capture (‘capture touch that the victims entered’).
  • [T1548 ] Abuse Elevation Control Mechanism – It abuses Wireless Debugging/ADB and Device Administrator prompts to gain higher privileges (‘get shell user capabilities’, ‘grant Device Administrator privileges’).
  • [T1021.006 ] Remote Services: Windows Remote Management – Not applicable; no matching Windows remote service technique is described.
  • [T1090 ] Proxy – The malware uses a persistent WebSocket channel to relay commands and data between device and C2 (‘persistent WebSocket connection’).
  • [T1036 ] Masquerading – It disguises malicious activity with fake installer screens, fake system updates, and spoofed login/lock screens (‘fake installation interface’, ‘deceptive “system update”‘).
  • [T1566 ] Phishing – It uses HTML overlays that imitate legitimate banking and transaction screens to trick victims into entering credentials (‘replicate the appearance of the genuine application’s login’).
  • [T1140 ] Deobfuscate/Decode Files or Information – The dropper decrypts and extracts the payload from the app’s assets folder (‘decrypts and extracts the actual payload’).
  • [T1053 ] Scheduled Task/Job – Not applicable; no task scheduling behavior is described in the article.
  • [T1556.002 ] Modify Authentication Process: Password Filter DLL – Not applicable; no password filter mechanism is described.
  • [T1562.001 ] Impair Defenses: Disable or Modify Tools – It blocks communication with Google Play/Google Play Services and bypasses background restrictions and OEM power controls (‘block network communication’, ‘neutralize OS background restrictions’).
  • [T1406 ] Input Capture – It uses overlays and Accessibility Service to intercept entered PINs and credentials (‘PIN harvesting process’, ‘capture touch’).
  • [T1647 ] Web Service – The malware loads attacker-controlled content inside a full-screen WebView for phishing (‘remotely loads an attacker-controlled URL’).

Indicators of Compromise

  • [Domain / URL ] C2 and malicious content delivery infrastructure – Amazon AWS-hosted buckets, attacker-controlled URL
  • [Network Protocol / Port ] Remote command-and-control and pairing channels – HTTPS, WebSocket, 127.0.0.1, and other dynamic pairing port
  • [File / Path ] Malware payload and fake content sources – application assets folder, resources/, WebView-loaded HTML templates
  • [Application / Package Targets ] Financial app targeting lists – 349 banking/financial/e-wallet/cryptocurrency applications, 140+ banking and crypto apps
  • [String / Command Names ] Malware feature and control identifiers – v2, replacePinTargets, catAllViewSwitch, admSet, admPwd, transparent, autoBoot
  • [IP Address ] Local ADB pairing endpoint – 127.0.0.1
  • [Crypto / Auth Data ] Wireless Debugging pairing artifacts – 6-digit PIN, SPAKE2 pairing code, dynamic connection port


Read more: https://zimperium.com/blog/the-toxicpanda-never-sleeps-toxicpanda-2.0-prepares-its-next-strike-on-mobile