A recent report by Glow has uncovered that AI coding agents, when tasked with sharing screenshots of code changes, inadvertently uploaded over 13,000 internal images from more than 300 organizations to public GitHub repositories. This sensitive data included customer billing records and unreleased feature designs, primarily stored under developers' personal accounts. The incident raises significant concerns regarding the oversight and security protocols surrounding AI tools in software development environments.
For businesses utilizing AI coding agents, this incident underscores the critical need for robust data governance practices. Organizations must implement stringent access controls and monitoring mechanisms to prevent sensitive information from being inadvertently exposed through AI interactions. This situation highlights the importance of training developers on the secure use of AI tools and the necessity of incorporating security measures into the development lifecycle. As the reliance on AI technologies grows, so too does the responsibility to safeguard proprietary and sensitive information, making cybersecurity a paramount consideration in the adoption of AI solutions.
---
*Originally reported by [The Hacker News](https://thehackernews.com/2026/09/ai-coding-agents-exposed-13000-internal.html)*