Building Aigris: An On-Device AI Cybersecurity Agent for Android

ullah-naqeeb — Wed, 08 Jul 2026 11:04:06 +0000

Cyber threats are becoming increasingly common. Phishing websites, malicious links, fake messages, and social engineering attacks target millions of people every day.

Most security solutions rely heavily on cloud services: data is sent to remote servers, analyzed, and then a decision is returned to the user.

As someone who is passionate about both artificial intelligence and cybersecurity, I started asking myself a question:

Can we build a security assistant that uses AI directly on the user's smartphone while preserving privacy?

This question became the starting point of Aigris.

Who am I?

I'm a second-year Computer Engineering student with a strong interest in:

Artificial Intelligence
Cybersecurity

I enjoy building projects that allow me to combine theory with practical engineering challenges. Aigris is currently my biggest personal project and an opportunity to explore how AI and cybersecurity can work together in real-world applications.

What is Aigris?

Aigris is an experimental Android application that aims to become an AI-powered cybersecurity agent capable of helping users defend themselves against common digital threats.

The project focuses on a simple principle:

Security should be intelligent, proactive, and privacy-friendly.

The long-term vision is to create a mobile security assistant capable of:

Detecting phishing attempts
Analyzing suspicious URLs
Identifying potentially dangerous content
Warning users about digital threats in real time
Performing as much analysis as possible directly on the device

Why On-Device AI?

Privacy is one of the main motivations behind this project.

Sending sensitive information to external servers is not always desirable. Modern smartphones are becoming increasingly powerful and can now execute lightweight machine learning models efficiently.

Running AI directly on the device offers several advantages:

Better privacy
Lower latency
Reduced dependence on cloud services
Some protection capabilities even while offline
Greater user control over personal data

This privacy-first approach is one of the core ideas behind Aigris.

Technologies Behind the Project

Aigris combines multiple areas of computer science:

Android Development
Kotlin
Machine Learning
ONNX Runtime
Cybersecurity
Software Engineering

One of the most interesting aspects of this project is learning how to deploy AI models efficiently on mobile devices while respecting performance and battery constraints.

Current Goals

The project is still evolving, but the current areas of focus include:

Phishing detection
Suspicious URL analysis
Real-time threat alerts
Privacy-preserving AI inference
Malware-related threat detection

Technical Challenges

Building an intelligent security application on a smartphone is not easy.

Some of the challenges I am currently exploring include:

Optimizing AI models for mobile devices
Reducing inference time
Limiting memory consumption
Balancing security and battery usage
Designing a user experience that remains simple and non-intrusive

These challenges are exactly what make the project exciting from both an engineering and research perspective.

Future Ideas

Some ideas I would like to explore in future versions are:

More advanced phishing classifiers
Social engineering detection
Better explainability of AI decisions
Browser protection mechanisms
Additional security modules
Further optimization of on-device models

Why I'm Sharing This

One of the reasons I love open source is the opportunity to learn from other developers.

I'm still a student, and I know there is a lot I can improve. That's exactly why I'm sharing this project publicly.

I would genuinely appreciate feedback about:

The architecture
The idea itself
The machine learning approach
The Android implementation
Possible security improvements
Features that could make the project more useful

GitHub Repository

You can find the project here:

https://github.com/ullah-naqeeb/aigris.git

If you have suggestions, ideas, or constructive criticism, I would be extremely grateful to hear your thoughts.

Even a small piece of feedback can help improve the project and help me grow as a developer.

Thank you for reading!

DEV Community: ullah-naqeeb