Protecting the device we rely on most: new AI project tackles rising mobile cyber threats

A new research project led by experts at the University of Wolverhampton (UoW) and University of Lancashire (UoL) is tackling one of the fastest‑growing digital risks of our time: cyberattacks targeting smartphones and tablets.

AgenticDeviceShield is an innovative project developing a new form of cyber defence that lives directly on mobile devices, protecting users from increasingly sophisticated scams, spyware and digital fraud – without sending personal data to the cloud.

The project is led by Dr Nazmul Hussain, PhD (UoL), alongside Professor Md Arafatur Rahman (UoW), and brings together expertise in artificial intelligence (AI), cybersecurity and human‑centred computing.

Why mobile security matters

Smartphones and tablets are now essential to everyday life. For billions of people worldwide, they are the main (and often only) way to access banking, healthcare, education and government services. By 2026, smartphones are expected to become the dominant way people access online services globally.

While mobile operating systems have become more secure at a technical level, cybercriminals have adapted. Rather than attacking the internal software of a device, many modern attacks focus on tricking the user.

These include:

• Phishing scams - where fake messages imitate trusted organisations such as banks or doctors
• AI‑powered social engineering - where messages are personalised using artificial intelligence to appear more convincing
• Deepfakes - where realistic audio or video is used to impersonate trusted individuals
• Malicious apps - installed outside official app stores
• ‘Zero‑click’ attacks - where a device can be compromised without the user clicking anything at all

As a result, smartphones can be technically secure yet still highly vulnerable.

In 2024 alone, more than 33 million mobile malware and phishing incidents were recorded worldwide, and mobile devices are now involved in over half of enterprise cyber incidents. The impact falls hardest on older adults, people with limited digital confidence, and users who rely entirely on a single device for essential services. For these individuals, one successful attack can lead to financial loss, identity theft, or exclusion from healthcare and public services.

Limitations of current solutions

Most mobile security tools today rely on:

• Cloud‑based systems, which require constant internet connectivity
• Signature‑based detection, meaning they only recognise known threats
• Opaque alerts, which give little explanation to the user

This means protection can fail when a device is offline or in areas with poor signal, and users are often left confused by warnings they do not understand – reducing trust and increasing risk.

How AgenticDeviceShield works

AgenticDeviceShield takes a fundamentally different approach.

The system uses agentic AI, meaning an AI system that can act independently, make decisions and respond to threats in real time. Crucially, it runs entirely on the device itself, rather than in the cloud.

Key features include:

• On‑device monitoring – the system continuously watches for suspicious behaviour without sending personal data anywhere
• Autonomous defence – it can respond to threats immediately, without waiting for updates
• Knowledge‑augmented AI – the AI combines learned patterns with structured cybersecurity knowledge, improving accuracy and reliability
• Offline resilience – protection continues even without an internet connection
• Explainable alerts – instead of technical warnings, users receive clear, plain‑language explanations of what is happening and what action is being taken

This means users are not only protected but also empowered to understand digital risks – regardless of their technical background.

A step towards safer, more inclusive digital access

By embedding privacy‑preserving, intelligent security directly into smartphones and tablets, AgenticDeviceShield aims to reduce digital harm and support safer access to essential online services.

The project reflects a growing recognition that cybersecurity is not just a technical issue, but a societal one, requiring solutions that are trustworthy, transparent and designed around real human needs.

Professor Md Arafatur Rahman said:
“AgenticDeviceShield represents a shift towards security systems that can reason, explain and act locally – creating protection that is both intelligent and humane.”

Dr Nazmul Hussain said:
“Mobile devices have become people’s lifeline to the digital world. Our goal is to make that access safer without requiring users to be cybersecurity experts or to sacrifice their privacy.”

The project places the Universities at the forefront of research into next‑generation mobile cyber defence, with potential long‑term benefits for individuals, businesses and public services alike.