Exploiting Android accessibility services

Issue 7 2022 Information Security

Android and iOS accessibility features are available to help people more easily use their smartphones, and include features such as audio comments, subtitles, custom display and so on. Some mobile applications designed with an inclusive approach are compatible with these accessibility services.

To enable these services in an application, it requires the accessibility permission. But this permission gives applications full access to the user’s device. Today, more cybercriminals are using this option to take control of smartphones and tablets. When this happens, users find themselves in a bind, unable to uninstall the app or even reset their device.

Recently, Pradeo Security neutralised an application using Android accessibility services for malicious purposes on a protected device. The identified malware was installed through a phishing link. It pretends to be a QR code scanning application but actually exploits the accessibility permission to perform fraudulent banking transactions.

The risks of mobile accessibility services

An application can use the BIND_ACCESSIBILITY_SERVICE permission in order to benefit from advanced features facilitating accessibility to users with disabilities. With this permission, an application can control the whole screen (clicks, movements, etc.) as well as the keyboard, read what is displayed, and close or open applications.

These features are sensitive because they enable the control of almost all layers of a device. When a malicious application is granted the accessibility permission, it can send all the information displayed on the screen and typed on the keyboard to a remote server, prevent its own removal or a system reset, and even launch itself automatically when the device is rebooted. Unfortunately, the distribution channels used by hackers, such as unofficial application stores and messaging services (SMS), do not provide any protection against this threat.

Case study: QR-Code Scanner

Name of the analysed app: QR-Code Scanner

Package name: com.square.boss

OS: Android

The ‘QR-Code Scanner’ application appears as a QR code scanning application. Its icon and name are not suspicious. However, when launched, no QR code scanning functionality is offered.

Immediately, the application sends a notification that urges the user to grant the accessibility option, which is necessary for the execution of its attack. As long as the user does not allow it, it continuously sends the same permission request.

Once authorised, the malware can silently approve its own permission requests in place of the user. Thus, it grants itself all the permissions that will allow it to carry out its attack.

In this case, our analysis of the malware suggests that the goal of the hacker behind the application is to commit fraud, by collecting data the user types or displays on their screen (login, password, credit card numbers, etc.) and intercepting temporary authentication codes that get sent.

First, the QR-Code Scanner application accesses the list of applications installed on the victim’s device to gauge interest. When banking or e-commerce applications are used, there is a greater chance that banking data is entered by the user. When it happens, the hacker collects them.

To enter the victim’s account or make a payment with their credit card, the hacker intercepts the one-time password contained in an SMS or a notification. Hence, they bypass all security measures that authenticate payments and connections using a code. Only verification protocols that use biometric data are safe at this point.

Finally, the application uses the victim’s phone to spread to other devices. To do this, it sends an SMS containing a phishing link to the entire contact list. This way, the message comes from a known number and has a better chance of convincing the recipients to install the malware.

Throughout the attack, the malware exploits accessibility services to:

• Spy on user activity.

• Grant and prevent the rejection of the permissions it needs.

• Prevent removal of the application, either from the homepage or from the settings.

• Prevent factory reset, even from a third-party device.

• Prevent sleep or shutdown of its process.

• Launch at startup.

The permissions used by the malware are the following:

• android.permission.QUERY_ALL_PACKAGES

• android.permission.QUICKBOOT_POWERON

• android.permission.RECEIVE_LAUNCH_BROADCASTS

• android.permission.GET_TASKS

• android.permission.SYSTEM_ALERT_WINDOW

• android.permission.RECEIVE_SMS

• android.permission.READ_SMS

• android.permission.WRITE_SMS

• android.permission.SEND_SMS

• android.permission.REQUEST_IGNORE_BATTERY_OPTIMIZATIONS

• android.intent.action.BOOT_COMPLETED

• com.htc.intent.action.

QUICKBOOT_POWERON

• android.intent.action.

QUICKBOOT_POWERON

• android.permission.

RECEIVE_BOOT_COMPLETED

• android.permission.QUICKBOOT_POWERON

Protective measures

Despite the undeniable need for accessibility services, the advanced rights they offer on the system mean they must be used (on the developer side) and authorised (on the user side) with due consideration. Today, only a few tools and remediation actions are effective at neutralising the malware:

• Blocking the application before launching it.

• Forcing the uninstallation of the application.

• Uninstalling via a device management solution (UEM, MDM).

• Uninstalling via ADB command.

Find out more at www.pradeo.com




Share this article:
Share via emailShare via LinkedInPrint this page



Further reading:

Compressing cyberattack timelines and targeting ungoverned AI identities
Information Security News & Events
Sophos AI Security 2026 report finds attackers are moving beyond experimentation and operationalising AI for attacks using identity as the primary initial access vector, rather than inventing new attack types.

Read more...
Zero-touch automation certificate life cycle management loop
Products & Solutions Information Security Security Services & Risk Management
ManageEngine completes the certificate life cycle management loop with CA-agnostic, zero-touch automation. New post-deployment automation in Key Manager Plus removes the last manual step in certificate renewal as lifespans gradually shrink to 47 days

Read more...
Sophos launches AI-native cybersecurity defence system
News & Events Information Security Security Services & Risk Management
Built for a threat landscape reshaped by AI, Sophos Fusion unites security operations, endpoint, network security, identity, email, and cloud into one defence system that prevents, detects, investigates, and responds at AI speed.

Read more...
Ungoverned AI agents and deepfakes pose critical threats
Information Security Security Services & Risk Management
Global study reveals 64% of South African organisations already deploy autonomous AI agents with little to no governance, while 63% of employees admit they are unlikely to be able to spot attacks such as deepfakes

Read more...
How ‘TikTok Brain’ is breaking legacy security training
Training & Education Information Security
Between doomscrolling, rapid-fire Slack notifications, and algorithmic video feeds, the average employee is trapped in an aggressive, highly engineered dopamine loop that automatically shuts down in traditional training situations.

Read more...
Quantum is coming
Infrastructure Information Security
The global cybersecurity landscape is approaching a turning point as quantum computing accelerates faster than most organisations realise; the shift is not a distant, theoretical concern, but a present-day business risk that demands immediate action.

Read more...
Outpacing cyberthreats in the age of AI
SMART Security Solutions Technews Publishing News & Events Information Security
SMARTpod talks to Fred Streefland, Global Field CISO for EMEA at Check Point Software Technologies, about framing modern cyber defence around adaptability, rapid decision-making, and the OODA loop adapted for cybersecurity.

Read more...
Prompt injection is the new phishing
Information Security Security Services & Risk Management
AI security is heading in an uncomfortable direction following Microsoft’s research showing how prompt injection can be chained to remote code execution vulnerabilities in AI agent frameworks, including work involving Semantic Kernel

Read more...
Cybersecurity needs actual intelligence before artificial intelligence
Information Security AI & Data Analytics
Cybersecurity depends on interpretation. A tool can tell you that something unusual has happened, but people need to determine whether it is a genuine risk, the business impact, and how to respond without causing unnecessary disruption.

Read more...
Duxbury Cybersecurity sharpens reseller offering
Duxbury Networking Information Security News & Events
Duxbury Networking has strengthened its Duxbury Cybersecurity business unit by adding WatchGuard and Cynet, giving South African resellers broader, more integrated coverage for the security risks customers are now asking them to address.

Read more...










While every effort has been made to ensure the accuracy of the information contained herein, the publisher and its agents cannot be held responsible for any errors contained, or any loss incurred as a result. Articles published do not necessarily reflect the views of the publishers. The editor reserves the right to alter or cut copy. Articles submitted are deemed to have been cleared for publication. Advertisements and company contact details are published as provided by the advertiser. Technews Publishing (Pty) Ltd cannot be held responsible for the accuracy or veracity of supplied material.




© Technews Publishing (Pty) Ltd. | All Rights Reserved.