RatHat Android trojan uses AI to steal banking data

A newly discovered Android banking trojan dubbed RatHat is using generative AI to navigate compromised devices and steal financial data, raising fresh concerns about how attackers are weaponizing consumer AI assistants for hands-free account takeover^[2][8][13]. The malware was detailed this week by Zimperium’s zLabs team, which describes RatHat as an AI-powered mobile threat that abuses Android’s Accessibility features and debug capabilities to maintain deep, remote control over infected phones^[8][3][11].

RatHat is delivered as a malicious APK that victims are tricked into installing, often through social engineering that encourages users to sideload what appear to be legitimate banking, cryptocurrency or utility apps^[3][9][15]. Once installed, the trojan requests extensive Accessibility permissions, uses overlay screens to capture banking logins, and intercepts SMS messages and one-time passcodes to defeat multi-factor authentication^[8][9][11]. Researchers report that RatHat also abuses Android’s Wireless Debugging feature to pair with the device’s local Android Debug Bridge (ADB), giving operators shell-level access that can be used to reinstall the malicious app even after a victim believes it has been removed^[3][10][12].

What sets RatHat apart from earlier mobile banking malware is its AI-driven user interface automation engine, which serializes the device’s live Accessibility tree into XML and sends it to a widely used generative AI assistant for analysis^[2][4][8]. The AI service, contacted in Mandarin according to multiple analyses, responds with element coordinates, on-screen text and navigation instructions such as SCROLL_DOWN, allowing the malware to decide in real time where to tap, scroll or input data^[3][4][13]. By delegating navigation decisions to a general-purpose AI assistant rather than hardcoded scripts, RatHat can adapt to different screen sizes, Android versions and localized app interfaces, making traditional signature-based detection and static UI automation far less effective^[1][2][6].

Zimperium’s threat summary and subsequent reporting describe RatHat as an active, globally distributed mobile banking trojan targeting financial, cryptocurrency and consumer banking applications with critical severity^[3][8][15]. The malware is engineered to harvest bank credentials, payment PINs, lock-screen PINs or patterns, and other sensitive data that can be used to drain accounts or take over devices^[1][3][9]. One investigation assesses the campaign as likely operated by China-based threat actors, though formal attribution has not yet been publicly confirmed by law enforcement or platform vendors^[10][7].

To reduce the chances of detection and analysis, RatHat incorporates multiple anti-analysis techniques, including inconsistent ZIP containers, padded Android manifests, malformed DEX pseudo-instructions and aggressive string obfuscation^[4][11][14]. The trojan can dynamically change its app icon and label to impersonate trusted applications, further tricking users into entering credentials or granting permissions they would normally reserve for legitimate banking or system apps^[11][14]. Combined with its AI-driven automation, these tactics position RatHat at the intersection of advanced mobile malware and emerging AI-enabled attack tooling, highlighting how quickly criminal operators are experimenting with consumer AI platforms to scale their operations^[2][8][13].

Because RatHat exploits platform features such as Accessibility and Wireless Debugging rather than a single patchable software flaw, no CVE has been assigned and there are no vendor security updates that can fully neutralize the threat on their own^[3][8][11]. Enterprise defenders are instead urged to restrict or monitor use of Wireless Debugging, lock down sideloading policies, closely scrutinize Accessibility permission requests, and deploy mobile threat defense tools with updated detections for RatHat’s behavior and infrastructure^[1][8][15]. End users can reduce their risk by avoiding sideloaded apps, reviewing permissions for installed software, and treating unexpected prompts for Accessibility access or debugging features as potential signs of compromise^[5][7].

References

1. New Android malware uses AI to steal bank logins and PINs
2. New RatHat Android malware uses AI to automate device …
3. RatHat Android Trojan Uses Generative AI for Device Takeover
4. RatHat: AI-Driven Android Malware Abuses Wireless Debugging to …
5. New Android Malware Uses AI to Steal Bank Logins and Reconstruct Your PIN
6. New RatHat Android malware uses AI to automate device …
7. New Android malware uses AI to take over your phone
8. RatHat: AI-Powered Mobile Threat is Here for Your Credentials …
9. New RatHat Android Malware Uses AI and ADB to Steal Banking Credentials and OTPs
10. RatHat Android Malware Abuses ADB to Retain Shell Access After Uninstall
11. Cyware Daily Threat Intelligence – September 17, 2026
12. RatHat Android Trojan Reinstalls Itself via Wireless ADB
13. New ‘RatHat’ Android Malware Leverages AI to Steal Financial Data
14. RatHat Turns Android Accessibility Into an Attack Weapon
15. New RatHat Android Malware Uses AI to Steal Banking Logins …