AI Model Achieves 98.53% Accuracy In Detecting Ransomware On Smart Devices

Scientists developed an AI model detecting ransomware in IoT devices with high accuracy, using deep learning and optimization techniques for cybersecurity.

A team of researchers has detailed their findings today in a Scientific Reports paper published by Nature, introducing an advanced AI-powered model designed to detect and prevent ransomware attacks on smart devices.

With the rapid expansion of Internet of Things (IoT) technology in homes, healthcare, and industries, cybersecurity threats have become a growing concern.

Ransomware, one of the most dangerous cyber threats, locks users out of their systems until they pay a ransom. The researchers explained how traditional security measures often fail to detect and prevent these evolving attacks, prompting researchers to explore AI solutions.

Their newly developed model, called Multi-head Attention-Based Recurrent Neural Network with Enhanced Gorilla Troops Optimization (MHARNN-EGTOCRD), significantly improves ransomware detection accuracy using machine learning techniques.

The model first normalizes incoming data using min-max normalization, ensuring efficient processing. It then employs Dung Beetle Optimization (DBO)—inspired by how dung beetles locate food—to filter out unnecessary information, focusing only on the most relevant cybersecurity threats.

At its core, the system utilizes a Multi-head Attention and Long Short-Term Memory (MHA-LSTM) network, an advanced deep learning approach that helps detect complex attack patterns.

By analyzing past ransomware behaviors, the AI can predict and flag potential attacks before they fully execute. Additionally, the system is fine-tuned using Enhanced Gorilla Troops Optimization (EGTO), which optimizes the AI’s settings for maximum efficiency.

In testing, the model achieved an impressive 98.53% accuracy in detecting ransomware, outperforming traditional cybersecurity methods. This high precision suggests that AI could become a powerful tool in the fight against cybercrime, particularly in safeguarding smart devices from sophisticated attacks.

The researchers believe their model could be integrated into existing cybersecurity systems, providing an early warning mechanism for ransomware attacks.

As IoT devices continue to expand in everyday life, strengthening their security is crucial to preventing financial and data losses. By combining nature-inspired optimization techniques with deep learning, this AI model represents a significant step forward in cybersecurity.