AI-Powered Data Leak Detection

I led the development of Cycode’s AI Leak Analyser—a system that finds sensitive organizational data exposed in public repositories.

Instead of flagging every mention of a company domain, it uses contextual AI to tell a real exposure (credentials, internal URLs, employee emails) from a benign reference. The result: analysts review 2–5 real leaks a day instead of hundreds, with the reasoning attached to each alert.

Read the write-up.