Critical Code Execution Vulnerability in Python Pillow Library via Crafted EPS Images

A critical vulnerability identified as CVE-2026-112900 has been discovered in the Python Pillow library, a widely adopted open-source image processing tool used across web frameworks, content management systems, and data science pipelines. The flaw resides in the library's handling of Encapsulated PostScript (EPS) image files.

According to the disclosure, when Pillow encounters an EPS file, it invokes Ghostscript—a separate program responsible for rendering the image—to convert the file into pixels. The vulnerability arises because Pillow passes the EPS filename to Ghostscript without adequate sanitization. An attacker can craft a malicious EPS image containing embedded command sequences. When a targeted application processes this file, the unsanitized filename is relayed to Ghostscript, allowing command injection and subsequent arbitrary code execution.

The impact is assessed as high due to Pillow's extensive integration. Numerous web frameworks and CMS platforms utilize Pillow for image upload processing, resizing, and thumbnail generation. If these systems accept user-uploaded images without strict validation, an attacker could exploit this vector to achieve remote code execution under the privileges of the application.

The Python Software Foundation and the Pillow Team have addressed the issue by releasing security updates. Users are advised to update Pillow to the latest patched version. The release notes indicate that version 10.0.1 or later contains the fix, though users on specific release series should verify the appropriate patched version for their current installation. In addition to updating, security recommendations include implementing strict input validation for uploaded files and reviewing image processing code paths that handle external filenames.