CVE-2026-46520

What it actually does

CVE-2026-46520 is an out-of-bounds heap write in ImageMagick that can be triggered when the library reads multiple images with different dimensions. A crafted multi-image input, or a workflow that processes multiple attacker-controlled images together, can cause ImageMagick to write past the bounds of an allocated heap buffer. The published impact is availability-focused: exploitation can crash the ImageMagick process or a dependent service that invokes ImageMagick. Because this is memory corruption in native code, organizations should treat it as potentially higher risk in exposed image-processing pipelines even though no public exploitation maturity is currently reported.

Why it matters to the business

Systems that accept, transform, thumbnail, resize, preview, or validate uploaded images may be disrupted by specially crafted image files. The most likely business consequence is denial of service for web applications, media-processing workers, content management systems, document workflows, or background job queues that rely on ImageMagick. ImageMagick is widely embedded in downstream software and automation, so exposure may exist indirectly through applications, container images, language bindings, and packaged appliances rather than only through manually installed ImageMagick binaries. The vulnerability is not currently listed in CISA KEV and has a low EPSS score, but services processing untrusted images should still prioritize patching because the attack requires no authentication where image upload or remote image ingestion is exposed.

How an attack unfolds

An attacker first identifies an application, API, email/document pipeline, content management feature, or batch-processing service that uses ImageMagick to read multiple images or multi-frame image inputs. The attacker then supplies crafted image content with inconsistent dimensions designed to trigger the heap out-of-bounds write during parsing or conversion. When ImageMagick processes the malicious input, memory corruption occurs in the ImageMagick process or in the calling application process, depending on how the library is integrated. In the expected scenario, this causes the image conversion, thumbnailing, or preview generation job to crash, potentially leading to repeated worker failures, service instability, queue buildup, or denial of service if the payload is repeatedly submitted. There are no named threat actors or known ransomware campaigns associated with this CVE based on the provided data.

How to fix it

Upgrade ImageMagick to a patched release: version 6.9.13-48 or later for the 6.x branch, or version 7.1.2-23 or later for the 7.x branch. Update both operating-system packages and any bundled copies in containers, appliances, application dependencies, CI/CD images, serverless layers, and language-specific wrappers such as libraries that invoke ImageMagick. Rebuild and redeploy affected images after upgrading. If immediate patching is not possible, reduce exposure by blocking untrusted multi-image or multi-frame formats where feasible, enforcing strict file type allowlists, applying image size and frame count limits, disabling remote URL image fetching, running ImageMagick in a sandboxed low-privilege environment, and isolating image-processing workers from critical application components. Configure resource limits using ImageMagick policy controls and operating-system controls to reduce denial-of-service impact.

How to confirm the fix worked

Confirm the installed ImageMagick version with commands such as `magick -version`, `convert -version`, or package-manager inventory and verify it is at least 6.9.13-48 for ImageMagick 6 or 7.1.2-23 for ImageMagick 7. Check container image manifests, SBOMs, software composition analysis results, and runtime hosts for older ImageMagick packages or embedded binaries. Validate that applications and workers are actually using the upgraded binary or library path after deployment. Test representative image upload, conversion, thumbnailing, and preview workflows after the upgrade to ensure normal functionality. Where compensating controls are used, verify that unsupported formats, excessive dimensions, excessive frame counts, and remote image fetches are blocked or constrained before ImageMagick processes the content.

Typical attack path

1. Find an externally reachable feature or internal ingestion workflow that accepts images or documents and uses ImageMagick for decoding, conversion, thumbnailing, or preview generation.
2. Create or obtain a malformed multi-image or multi-frame input that contains images with differing dimensions and triggers the vulnerable read path.
3. Submit the crafted file through an upload form, API endpoint, message attachment, document conversion workflow, or remote URL ingestion feature.
4. ImageMagick processes the attacker-controlled input and performs an out-of-bounds heap write while reading the images.
5. The ImageMagick command, worker process, or dependent service crashes or becomes unstable, causing failed conversions, repeated job retries, resource exhaustion, or denial of service.

How to detect it

To determine whether exploitation was attempted or successful, review logs from applications that accept uploads or process images, ImageMagick wrapper logs, web access logs, API gateway logs, background worker logs, container logs, system journals, crash reporters, EDR telemetry, and core dump repositories. Look for ImageMagick-related process crashes, segmentation faults, heap corruption messages, repeated conversion failures, worker restarts, or job retries immediately after image uploads or remote image fetches. Correlate suspicious events with submitted files, source IPs, user agents, API tokens, queue message IDs, and timestamps. Pay particular attention to multi-image or multi-frame inputs and files with unusual dimensions, inconsistent frame sizes, excessive frame counts, uncommon image formats, or malformed headers. In Linux environments, search for kernel or systemd messages showing `magick`, `convert`, or application worker crashes with SIGSEGV, SIGABRT, malloc corruption, or heap-buffer-overflow diagnostics. In containerized environments, look for repeated pod restarts, exit codes associated with crashes, and logs from image-processing jobs that fail on specific files. Preserve suspicious image samples and core dumps for controlled offline analysis, but do not reprocess samples on production systems.

• magick: Segmentation fault
• convert: Segmentation fault
• magick: Aborted
• convert: Aborted
• heap-buffer-overflow
• out-of-bounds write
• AddressSanitizer
• malloc(): corrupted top size
• malloc(): invalid size
• double free or corruption
• corrupted double-linked list
• free(): invalid pointer
• SIGSEGV
• SIGABRT
• Process: magick
• Process: convert
• /usr/bin/magick
• /usr/bin/convert
• /usr/local/bin/magick
• /usr/local/bin/convert
• core.magick
• core.convert

What your security team should do

• Inventory all direct and indirect ImageMagick usage, including web upload handlers, CMS platforms, document conversion services, media workers, containers, build images, and language bindings.
• Prioritize patching internet-facing or unauthenticated image-processing paths to ImageMagick 6.9.13-48 or later, or 7.1.2-23 or later.
• Add or review upload controls: allow only required image formats, reject malformed files, enforce maximum dimensions and frame counts, and disable remote URL ingestion unless explicitly required.
• Run ImageMagick under least privilege with sandboxing, filesystem restrictions, no unnecessary network access, and CPU, memory, file, and process limits.
• Review recent crashes, worker restarts, failed conversion spikes, and unusual uploads for signs of attempted exploitation since publication on 2026-06-10.
• Update vulnerability management and SBOM tooling to detect vulnerable ImageMagick versions in host packages, containers, appliances, and application bundles.
• Because this is not currently KEV-listed and EPSS is low, use exposure-based prioritization, but do not defer remediation on systems that process untrusted content.