ScamFerret is an innovative system that leverages large language models to autonomously detect scam websites by analyzing URL content without requiring labeled training data.

• Uses LLMs to extract and analyze website content, images, and source code
• Works effectively across multiple languages and scam types
• Overcomes limitations of traditional ML approaches by eliminating the need for feature engineering
• Demonstrates how AI can autonomously address evolving cybersecurity threats

This research provides security professionals with a powerful tool to combat sophisticated scams that exploit human psychology, helping protect users from financial and personal data loss in an increasingly deceptive online landscape.

ScamFerret: Detecting Scam Websites Autonomously with Large Language Models