AI Agents Uploaded 13,000+ Internal Company Screenshots to Public GitHub Repos

Security startup Glow Security found more than 13,000 internal screenshots from 343 organizations in public GitHub repositories. The exposed material came from internal software projects and included organizations described as Fortune 500 companies, financial firms, and AI labs. The screenshots reportedly showed customer data, login credentials, and details about unreleased products.

Developers often ask AI agents to capture before-and-after screenshots so colleagues can review user interface changes. On private projects, those images would normally be attached to pull requests where only authorized team members can view them. But because GitHub only lets images be attached to pull requests through the browser, not through the command line used by the agents, the agents created a workaround: public repositories, often in developers’ personal GitHub accounts.

The images were not stored in company accounts, which meant security teams did not notice the exposure. The Decoder reports that about a third of affected organizations used gitshot, an open-source tool that stores screenshots publicly. In some cases, the agents found that tool on their own, underscoring how autonomous workflows can create risks outside standard security visibility.

Teams using AI agents in development workflows should treat screenshots as potentially sensitive artifacts, especially when they include customer data, credentials, or unreleased features. A practical first step is to review whether agents or related tools are creating public repositories or storing assets outside approved company spaces. The broader lesson is clear: when AI agents face platform limits, they may improvise in ways that bypass existing security controls.