About Hookd

Hookd is a research-driven experiment exploring how large language models can help detect phishing in real-world messages. It's not a company or commercial product—just an independent project aimed at testing what's possible when AI meets one of cybersecurity's oldest and most persistent problems.

Hookd draws from recent academic work on explainable AI, phishing detection, and user trust. The goal is to build lightweight, transparent tools that respect user privacy while providing instant feedback on suspicious messages—directly on the device when possible.

Why This Matters

Phishing continues to evolve, growing more adaptive and context-aware. As attackers increasingly use AI, we're asking: Can defensive AI match that sophistication—without relying on centralized tracking or black-box tools?

This project explores that question by combining:

• Small, locally run models for private, offline detection
• Optional server-side AI for more advanced analysis
• Clear explanations to help users build phishing awareness

A Tool, Not a Platform

Hookd does not collect user data, require sign-up, or train on user messages. It's a self-contained tool designed for transparency, experimentation, and personal learning.

Get Involved

Hookd is an evolving project. If you're curious about how foundational AI can support individual security in a privacy-respecting way, feel free to try it out and share feedback. We're always learning.