Detect SMS scams from screenshots using OCR and AI
Build a workflow that accepts an image upload of an SMS. Resize the image if needed, send it to OCRSpace to extract text, then pass the text to an LLM prompted to identify scam indicators (urgency, suspicious links, impersonation, payment requests). Return a verdict with the reasons. Handle low-quality images, OCR failures, and non-text images gracefully, and avoid storing personal data beyond what's needed. Output the extracted text and the scam-risk assessment.
What this prompt builds
Upload a screenshot of a text message and get an OCR + AI verdict on whether it's a scam.
The problem
Smishing (SMS phishing) is hard for users to judge and rarely reported in a usable form, so scams succeed and security teams have no signal. Pasting screenshots around for opinions doesn't scale.
Solution and impact
This workflow extracts text from an uploaded message screenshot with OCR and uses AI to flag scam indicators, returning clear feedback. Employees get an instant second opinion and the org gains visibility into smishing attempts.