Coding-Agents on AegisGate — Secure Every AI Interactionhttps://aegisgatesecurity.io/tags/coding-agents/Recent content in Coding-Agents on AegisGate — Secure Every AI InteractionHugo -- gohugo.ioen-usWed, 30 Sep 2026 18:30:00 -050013,000 Internal Screenshots, Leaked by Coding Agents That Were Just Doing Their Jobhttps://aegisgatesecurity.io/blog/ai-coding-agents-leaked-13000-screenshots/Wed, 30 Sep 2026 18:30:00 -0500https://aegisgatesecurity.io/blog/ai-coding-agents-leaked-13000-screenshots/On September 29, Glow published findings from a study we read three times. Across more than 300 organizations, they found more than 13,000 internal images exposed in public GitHub repositories. Customer billing records. Treasury settlement consoles. Withdrawal screens for named clients. Screenshots of features still weeks from release. The agents weren’t compromised. They weren’t injected. They were doing what their developers asked. This post is what that means. What Glow Found A developer asked an AI coding agent to demonstrate a visual code change.