Dev.to
7/19/2026

The original title is: "PhishVision: a Vision AI API for phishing detection and hidden prompt injection discovery"
Original: I Built a Free API That Detects Phishing Sites Using AI Vision — And It Catches Prompt Injection Too
Short summary
A solo developer built PhishVision, a free REST API that detects phishing sites by launching a headless Chromium browser, capturing screenshots, and sending them to Vision AI for forensic analysis. Unlike URL reputation databases, it catches brand-new phishing sites and hidden prompt injection attacks embedded in invisible page text. The API returns structured JSON verdicts with confidence scores, impersonated brand detection, and visual anomaly lists.
- •PhishVision uses headless Chromium + Vision AI to detect phishing sites that URL databases miss
- •Also detects hidden prompt injection attacks in invisible page text (white-on-white, display:none, 1px font)
- •Returns structured JSON verdicts with confidence scores and visual anomaly detection
Generated with AI, which can make mistakes.
Is this a good recommendation for you?



