Scientists have found a new way to spot hidden apps on smartphones, which could greatly assist law enforcement efforts. This research is detailed in the journal Future Internet.

This study, conducted by teams from Edith Cowan University and the University of Southern Queensland, demonstrates how machine learning can be effectively used to recognize these kinds of apps.

Smartphones are a crucial part of everyday life for about 5 billion people globally.

As concerns over digital privacy grow, more individuals are turning to vault apps. These applications offer secure storage by encrypting files and making them less visible in a phone’s system.

According to Associate Professor Michael Johnstone from ECU, while these apps can have beneficial uses, they are often linked to questionable activities.

“Vault apps allow users to conceal files, messages, and even entire applications with extra encryption layers,” said Associate Professor Johnstone. “This can be for legitimate reasons, like safeguarding personal photos or sensitive information.”

“However, they can also be misused for spying or other malicious activities that threaten user privacy and safety,” he explained.

Johnstone noted that these vault apps can be challenging to detect with current technology.

“Many of these apps mimic the look and functionality of standard applications, making them tricky to find,” he added. “Traditional detection methods often rely on knowing in advance which apps are problematic, complicating the process.”

This creates a significant obstacle for smartphone forensics, particularly for police investigations.

Researchers found that by employing machine learning, they could effectively identify vault apps, achieving high accuracy rates on Android devices.

“Our study suggests a more efficient method using machine learning, which doesn’t require a constantly updated list of problematic apps,” Associate Professor Johnstone stated. “We can detect an Android vault app with up to 98% accuracy.”

With such high accuracy, Johnstone believes this method could be extremely beneficial for law enforcement.

“This approach could play a vital role in helping address these important issues,” he said. “Given the widespread use of smartphones, any effective and non-intrusive method would be advantageous.”

The next steps in this research involve testing more algorithms, expanding the sample size, and exploring whether this approach could be adapted for non-Android devices, as well as understanding the use of vault apps by cybercriminals.