Accessibility-driven UI automation to self-grant permissions

Locates nodes in the Settings or package-installer user interface by text or view identifier and dispatches synthetic clicks. This lets the application approve its own permission dialogs, complete side-loading, and intercept attempts to reach the uninstall screen without any real user consent.

Boot-triggered restart with disguised foreground service

Restarts itself on device boot and holds a foreground service with a low-visibility notification channel, keeping the process alive across reboots and outside normal background limits. Malicious use is distinguished by pairing this with surveillance or remote-control primitives.

Covert microphone or camera capture with exfiltration

Opens audio or camera capture outside any user-facing recording feature and pairs it with file staging and network transmission. The absence of a foreground recording interface alongside these primitives indicates surveillance rather than a legitimate media function.

Keylogging via accessibility text-change events

Subscribes to text-change and view-focus accessibility events and reads the text of nodes belonging to other applications, capturing everything typed on the device including passwords entered into unrelated apps. Password-field flags are typically inspected so credential entry can be prioritised.

Notification listener abuse for OTP and message theft

Binds a notification listener service and reads posted notification content, optionally dismissing notifications afterwards to hide them. This harvests one-time passcodes and private messages without ever requesting SMS permissions, which is why it has largely replaced SMS interception in modern ban

Runtime string decryption with reflective method dispatch

Combines symmetric decryption or base64 decoding of embedded string tables with reflective class and method resolution, so sensitive API names never appear in plaintext. This defeats naive string-based detection and indicates deliberate anti-analysis engineering.