Detect image and QR-code phishing with ChatGPT Vision

Starting promptUse this prompt as a starting point to build your workflow.

Build a workflow triggered by a reported email. Render the HTML body to an image, then send the image to the OpenAI Vision API with a prompt to identify phishing indicators, including embedded QR codes and brand impersonation. Decode any QR code found and evaluate its destination. Handle rendering failures and large messages, avoid sending PII unnecessarily, and rate-limit API calls. Output the visual analysis verdict with the detected tactics.

New to Tines?Sign up free for Tines 3B Explore Edition

What this prompt builds

Render reported emails to images and use ChatGPT Vision to catch image- and QR-based phishing.

The problem

Modern phishing hides its payload in images and QR codes that text-based scanners miss entirely. These attacks sail through traditional filters and reach inboxes.

Solution and impact

This workflow converts a reported email's HTML to an image and analyzes it with ChatGPT Vision to spot visual phishing tactics like fake logos and malicious QR codes. Threats invisible to text analysis get caught and explained.