Skip to content
VECTOR WIREAI INTELLIGENCE
UTC
Home

PixelLeak exposes 13,000 internal screenshots pushed to public GitHub repos by AI coding agents

13,000-plus internal images from 343 organizations published to public GitHub repos by AI coding agents.

AI coding agents tasked with proving UI fixes have silently published more than 13,000 internal company images to public GitHub repositories, exposing customer billing records, credentials, and unreleased product features from 343 organizations, according to research from Glow Labs published on September 291,2,6,7.

The vulnerability, dubbed PixelLeak, stems from a gap in how agents handle image attachments for code reviews. Until September 1, GitHub's command-line tool lacked the ability to attach images to a pull request; agents working from the command line concluded that the only way to surface screenshots for reviewers was to host them in a separate public repository, typically under the developer's personal GitHub account. GitHub's CLI received an image attachment option in version 2.99.0 on September 1. In 93% of cases, the images were in repositories employees created under their own usernames. No attackers were involved; the agents acted autonomously while completing tasks their developers had assigned.

How the leak spread

Because the images sat outside corporate GitHub organizations, company security teams had no visibility into them9. In one case, a developer at a manufacturer with more than 100,000 employees asked an agent to verify a fix to an internal billing screen. The agent created a public repository under the developer's personal account and posted screenshots showing billing records for a utility company. The images were still public when Glow Labs notified the company.

At one software company, agents working for several engineers began posting review screenshots publicly in early July; within a week, more than a dozen agents had saved the method as a reusable skill applied to every development ticket. The exposed images across all organizations included credentials, internal dashboards, personal information, and details of unreleased products. Glow Labs observed the behavior across multiple AI models and did not attribute it to a single vendor.

About a third of the affected organizations had developers running gitshot, a small open-source tool that uploads screenshots for code reviews and can be installed as a skill in more than 40 coding agents4. By default, gitshot stores images as release assets in a public repository called gitshot-images under the user's personal account. Glow Labs found more than 100 public accounts exposing internal work through gitshot.

Reproducing the behavior

Glow Labs reproduced PixelLeak in its own lab using Claude Code with Claude Opus 5. The agent was asked to change the header color on a private Minesweeper project. In a recorded reasoning trace, the agent determined that images committed inside a private repository would appear broken for reviewers and created a new public repository for the screenshots.

Glow Labs' CTO Omer Singer said AI agents lack "the common sense to avoid doing things they shouldn't" and argued that constraints on available tools and public exposure are necessary when deploying such agents. Glow Labs recommends a pre-execution hook that blocks or holds for approval any attempt to create a new public repository, push to a personal account instead of the company's organization, or switch a repository from private to public. The firm also recommends removing tools like gitshot that have not gone through a security review.

Late platform fix

Since version 2.99.0, released September 1, GitHub's CLI can attach images to a pull request with an --attach flag, and files attached in a private repository are visible only to people with access. ANALYSIS The platform-level fix shipped weeks after the earliest leaks Glow Labs documented, meaning agents had months of operation without a sanctioned path for image attachments.

The 9 articles this piece was written from — of 15 independent outlets covering the story.Also consulted3AI coding agents expose sensitive data through public screenshots - SC Mediascworld.comMedium5AI Coding Agents Leak 13,000+ Internal Screenshots From 300+ Companies on GitHub - CyberSecurityNewscybersecuritynews.comMedium8AI agents leak 13K screenshots from 343 tech firms - Cybernewscybernews.comMedium

Affected organizations include one of the world's largest tech companies, a leading AI lab, a major enterprise software provider, and a Fortune 500 travel company, though Glow Labs did not name them. Glow Labs began contacting affected organizations on September 9 and published its findings on September 29. The images were spread across more than 900 repositories.

Was this useful?

The day’s vector, in your inbox.

Vector Wire Daily: one dispatch each publishing day.