CVE-2026-56365: ImageMagick PNG Encoder Memory Leak in MNG Processing

Technical summary

CVE-2026-56365 involves improper resource cleanup (CWE-401) in ImageMagick's PNG encoder implementation when writing MNG image formats. The vulnerability manifests as a memory leak triggered during encoder failure scenarios. When the PNG encoder encounters certain crafted MNG inputs, it fails to deallocate memory resources that were allocated during the encoding process. Repeated failures cause cumulative memory consumption until process limits are reached. The issue affects ImageMagick versions prior to 7.1.2-19, where memory cleanup logic was corrected. The attack vector is network-based with no authentication required, but a high complexity factor—specific image characteristics must be present to trigger the leak—limits practical exploitability.

Business impact

The primary business impact is service availability. ImageMagick is widely used in content management systems, web applications, image processing pipelines, and automated workflows. An attacker who can supply crafted MNG files to an exposed ImageMagick instance could degrade or disrupt image processing services. In multi-tenant or SaaS environments where ImageMagick processes user-uploaded content, this could be leveraged to create a denial-of-service condition affecting legitimate users. Organizations heavily reliant on uninterrupted image processing—such as media platforms, e-commerce sites with dynamic imaging, or document conversion services—face the greatest operational risk. Financial impact is limited to downtime and recovery costs rather than data breach or confidentiality loss.

Affected systems

ImageMagick versions before 7.1.2-19 are vulnerable. The issue is specific to the PNG encoder's MNG-writing code path, so systems must both run a vulnerable ImageMagick version and process MNG image files (or accept MNG inputs) to be at risk. Organizations using ImageMagick for general image manipulation may be unaffected if they do not encounter MNG inputs. However, any application that integrates ImageMagick for automated image processing, content transformation, or format conversion—especially those accepting user-supplied files—should be considered in scope. This includes web frameworks with built-in ImageMagick support, image optimization tools, and thumbnail generation services.

Exploitability

Exploitability is rated as low to moderate. While the attack vector is network-accessible and requires no authentication, the attack complexity is high. An attacker must craft a specific MNG image designed to trigger the encoder failure condition; generic or randomly malformed MNG files may not reliably reproduce the leak. The attacker also needs a way to deliver the malicious image to a vulnerable ImageMagick instance—typically by uploading to a web application or service that processes the file. In closed environments where image inputs are strictly controlled or validated, the practical risk is reduced. However, in open platforms accepting user uploads or processing untrusted image sources, the barrier to exploitation is lower.

Remediation

Upgrade ImageMagick to version 7.1.2-19 or later. This version corrects the memory leak in the PNG encoder's MNG handling. Organizations unable to upgrade immediately should implement input validation and filtering to reject or sanitize MNG files, implement resource limits and memory quotas on ImageMagick processes to prevent runaway consumption, and monitor for unusual memory growth patterns in image processing services. If feasible, restrict ImageMagick instances to trusted image sources and disable support for MNG format if it is not required for business operations.

Patch guidance

Verify the version of ImageMagick in use via 'convert -version' or equivalent. For systems running versions before 7.1.2-19, obtain the patched version from the official ImageMagick repository or your distribution's package manager. Test the patch in a non-production environment first to ensure compatibility with existing workflows and dependent applications. After patching, confirm memory behavior during typical workloads and stress-test with MNG files if MNG processing is part of your regular operations. For containerized deployments, rebuild container images with the patched ImageMagick version and redeploy.

Detection guidance

Monitor ImageMagick processes for sustained or rapid memory growth, particularly during image processing operations. Use process monitoring tools (top, htop, or cloud-native observability platforms) to establish baseline memory usage and alert on significant deviations. Collect and analyze ImageMagick error logs for repeated encoder failures or allocation errors. Implement file-type inspection at ingestion points to identify and log MNG file submissions. If possible, enable verbose logging in ImageMagick or use memory profiling tools (valgrind, AddressSanitizer if compiled in) in test environments to confirm leak patterns. Correlate memory exhaustion events with image upload or processing triggers to identify potential attack activity.

Why prioritize this

This vulnerability warrants medium priority for most organizations, despite its low CVSS score. While the attack requires complexity and delivers only denial-of-service impact, the widespread deployment of ImageMagick in production systems and the ease of delivering a malicious file make it a practical risk. Organizations with exposed image upload features, content transformation pipelines, or automated imaging workflows should prioritize patching. The lack of KEV/CISA awareness and absence of active exploitation reports suggest this is not yet a widespread target, providing a window to patch proactively before adoption increases. For organizations with strong availability requirements or multi-tenant architectures, earlier patching is justified.

Risk score, explained

The CVSS 3.1 score of 3.7 (LOW) reflects the attack vector (network-accessible, no authentication) and the denial-of-service outcome, but penalizes for high attack complexity and limited scope (no impact on confidentiality or integrity). This scoring is appropriate and conservative—it acknowledges that while the barrier to triggering the vulnerability is higher than simple network access, the impact on availability is real and can affect production systems. Organizations should not underestimate the score; even low-scoring availability-impacting vulnerabilities merit timely remediation, particularly in services with SLAs or critical business dependencies.

Frequently asked questions