Researchers unveil new way to counter mobile phone ‘account takeover’ attacks

Computer science researchers have developed a new way to identify security weaknesses that leave people vulnerable to account takeover attacks, where a hacker gains unauthorized access to online accounts.

Most mobiles are now home to a complex ecosystem of interconnected operating software and Apps, and as the connections between online services has increased, so have the possibilities for hackers to exploit the security weaknesses, often with disastrous consequences for their owner.

To understand and prevent these attacks, researchers had to get into the mind of the hacker, who can build a complex attack by combining smaller tactical steps.

Dr Luca Arnaboldi from Birmingham's School of Computer Science worked with Professor David Aspinall from the University of Edinburgh, Dr Christina Kolb from the University of Twente, and Dr Sasa Radomirovic from the University of Surrey to define a way of cataloguing security vulnerabilities and modelling account takeover attacks, by reducing them their constituent building blocks.

Until now, security vulnerabilities have been studied using ‘account access graphs’, which shows the phone, the SIM card, the Apps, and the security features that limit each stage of access.

However, account access graphs do not model account takeovers, where an attacker disconnects a device, or an App, from the account ecosystem by, for instance, by taking out the SIM card and putting it into a second phone. As SMS messages will be visible on the second phone, the attacker can then use SMS-driven password recovery methods.

The researchers overcame this obstacle by developing a new way to model how account access changes as devices, SIM cards, or Apps are disconnected from the account ecosystem.

Their method, which is based on the formal logic used by mathematicians and philosophers, captures the choices faced by a hacker who has access to the mobile phone and the PIN.

The researchers expect this approach, which is published in the Proceedings of the 28th European Symposium on Research in Computer Security (ESORICS 23), to be adopted device manufacturers and App developers who wish to catalogue vulnerabilities, and further their understanding of complex hacking attacks.

The published account also details how the researchers tested their approach against claims made in a report by Wall Street Journal, which speculated that an attack strategy used to access data and bank accounts on an iPhone could be replicated on Android, even though no such attacks were reported.

Apps for Android are installed from the Play Store, and installation requires a Google account, and the researchers found that this connection provides some protection against attacks. Their work also suggested a security fix for iPhone.

Apple has now implemented a fix for this, providing a new layer of protection for iPhone users¹.

The researchers repeated this exercise across other devices (Motorola G10 Android 11, Lenovo YT-X705F Android 10, Xiaomi Redmi Note Pro 10 Android 11, and Samsung Galaxy Tab S6 Lite Android). Here they found that the devices that had their own manufacturer accounts (Samsung and Xiaomi) had the same vulnerability as Apple – although the Google account remained safe, the bespoke accounts were compromised.

The researchers also used their method to test the security on their own mobile devices, with an unexpected result. One of them found that giving his wife access to a shared iCloud account had compromised his security – while his security measures were as secure as they could be, her chain of connections was not secure.

Dr Arnaboldi is currently engaged in Academic Consultancy where he works with major corporates and internet-based companies to improve their defences against hacking.