AI Phishing Detection & Social Engineering Lab
Analyze email headers, sender domains, embedded URLs, and deceptive psychological triggers to calculate objective phishing risk scores.
The Problem
What difficult challenge is this experiment tackling?Modern generative AI makes phishing emails nearly impossible for employees to spot, as attackers no longer make obvious grammatical mistakes.
The Hypothesis
What do I believe autonomous AI & engineering can accomplish?Analyzing homoglyph domain impersonation, SPF/DKIM/DMARC authentication, and NLP urgency patterns can expose sophisticated AI-crafted spear phishing.
System Architecture & Workflow
End-to-end design, data pipelines, and component interactionBenchmark & Empirical Metrics
Accuracy • Latency • Cost • Token Usage • Human Intervention98.5% detection accuracy on a test benchmark of 1,000 synthetic spear-phishing and legitimate transactional emails.
Security, Privacy & Ethical Guardrails
Risk mitigation, strict boundaries, and sandbox controlsEmails analyzed in isolated memory; sandboxed link inspection through safe headless Chromium container.
Lessons Learned & Observations
Empirical findings from prototyping and testingCombining technical header checks (DKIM pass/fail) with linguistic urgency detection yields the lowest false alarm rate.
Next Experiment Iteration
Upcoming modifications, scaling factors, or new hypothesesBuild a browser extension that warns users before entering credentials on lookalike domains.