ImageMagick EUVD-2026-97713

Exploitation requires ImageMagick to decode an attacker-supplied JPEG whose JPEG component count is fewer than four (e.g., single-component grayscale or 3-component YCbCr images), which is what leaves comp_info[] slots uninitialized when the pre-patch code formats all four. The realistic limiting factor is disclosure reachability: the uninitialized bytes are written into the image's sampling-factor property, so an attacker only gains value if the surrounding application exposes that metadata back to them - for example through 'identify'/property output, preserved metadata in a converted file, or an API that returns image attributes. On a pipeline that processes uploads but discards metadata, the leak is not observable. No authentication or user interaction is required at the ImageMagick layer (PR:N/UI:N) assuming an automated server-side processing path.

This is a genuine but low-to-moderate priority issue, not a high-urgency flaw. The CVSS 4.0 vector (AV:N/AC:L/AT:N/PR:N/UI:N) describes a remote, low-complexity, unauthenticated trigger, which is why the base score lands at 6.9; however, the impact metrics are the limiting factor - only VC:L (low confidentiality) with no integrity or availability impact (VI:N/VA:N). The leak is narrow: the JPEG decoder's JPEGSetImageSamplingFactor function unconditionally formatted four component sampling factors even when the image has fewer components (e.g., grayscale or 3-component YCbCr), reading uninitialized comp_info[] struct members and embedding them in the image's sampling-factor property. No EPSS score was provided, there is no CISA KEV listing, and no public exploit was identified - only vendor patch commits exist, which is distinct from weaponized exploit code. The practical risk hinges on whether an application returns the derived sampling-factor metadata back to an attacker; the base vector's UI:N/PR:N assumes an automated server-side processing pipeline. Net: a real defect worth patching on ubiquitous software, but the small, low-entropy disclosure makes it far less severe than the 'network/unauthenticated' metrics alone suggest.

An attacker uploads a crafted JPEG with fewer than four components to a web service that uses ImageMagick to process or convert the image. ImageMagick formats uninitialized comp_info memory into the sampling-factor property; if the application then exposes that metadata (e.g., via 'identify', an EXIF/property dump, or converted output), the attacker reads back small fragments of uninitialized process memory. No public exploit code was identified at time of analysis, though the upstream patch commits publicly reveal the root cause.