Students Detect Hidden Malware on Android with AI | Generative ai in investment management | Google generative ai course free | Generative ai in finance and accounting | Turtles AI

Researchers at Florida Polytechnic University have developed an AI model that can recognize remote access trojans (RATs) on Android with 99% accuracy. The technique, presented at the IEEE/ACIS conference, promises mobile tools to identify hidden malware.

Key Points:

• Specific focus on RATs, stealthy Android malware that is difficult to detect.
• Machine learning models trained on targeted samples.
• Detection rate approaching 99%.
• Next up: Mobile app and expansion to multiple malware families.

A team led by recent graduate Nesreen Dalhy (B.S.’23, M.S.’25), under the supervision of Dr. Karim Elish, has used AI to tackle one of the most insidious cyber threats: remote access trojans on Android devices. Using a strategy that included analysis of threat intelligence archives, the team built models that could capture behavioral details typical of RATs and distinguish SERP micro-samples between benign and malicious examples, something that conventional methods often fail to do. The result is a system with a detection accuracy approaching 99%, a significant figure that highlights the effectiveness of the specialized approach compared to generic detection solutions.

Dalhy highlighted how RATs – representing over 80% of mobile devices – can operate silently in the background, exploiting hidden permissions to perform invasive actions such as calls, audio/video recording or data exfiltration. Elish, an expert in Android security, specified that three distinct models were created capable of identifying “almost all” RATs with maximum effectiveness.

The research was formally presented during the IEEE/ACIS international conference on software engineering and cybersecurity applications, held in late May. The researchers plan to turn the prototype into a real mobile application in the near future and expand the algorithm to cover additional malware families, thus increasing the practical scope of the initiative.

The innovation reflects the real-world impact of student-led research in technology and security, proposing advanced solutions for the daily defense of millions of Android users.

This technology represents a significant step towards more targeted and dynamic protection against sophisticated malware.