CVE-2026-112403: Heap Buffer Overflow in Pillow's JPEG Draft-Scale Decoder
What's the Vulnerability?
CVE-2026-112403 is a heap-based buffer overflow (CWE-122) in the JPEG decoder of the Python Pillow imaging library. When an application uses Image.draft() or thumbnail-style decoding that requests a scaled-down JPEG, Pillow's decoder can write a full-width scanline into a buffer that was only sized for the smaller draft scale. That mismatch corrupts adjacent heap memory.
The impact is severe: an attacker who can get a crafted JPEG into an image-processing pipeline can trigger heap corruption, resulting in denial of service or, in the worst case, remote code execution. This vulnerability carries a CVSS score of 9.2.
Affected Software
- Vendor/Package: Pillow (Python Imaging Library), distributed via PyPI
- Affected versions: all releases prior to 13.0.0 (confirmed through 12.3.0)
- Fixed version: 13.0.0
Any application, service, or microservice that decodes untrusted JPEGs using Pillow's draft/scaling path — thumbnail generators, image upload pipelines, CDNs, document converters — is in scope.
Exploitation Status
A public proof-of-concept exists via the associated GHSA advisory. This is not yet confirmed as weaponized in the wild or added to a known exploited vulnerabilities (KEV) catalog, but the combination of a public PoC, a high CVSS score, and a straightforward trigger (a single crafted JPEG) means defenders should treat this as an active, near-term risk rather than a theoretical one. Attack surface is anywhere untrusted images are decoded with scaling.
How Our Detection Catches It
This detection ships across Microsoft Sentinel (KQL), Splunk (SPL), Elastic (EQL), QRadar (AQL), Sumo Logic, Chronicle (YARA-L), and CrowdStrike (CQL), and focuses on three signal categories:
- Vulnerable inventory: identifying hosts and services with a Pillow installation at or below the vulnerable version boundary (< 13.0.0), surfaced from package inventory and build/dependency telemetry.
- Crash correlation: flagging anomalous crashes, segfaults, or abnormal process termination in processes performing image decoding, which can indicate heap corruption from a malicious JPEG rather than an ordinary application fault.
- Ingestion anomalies: watching for unusual JPEG ingestion patterns against image-handling services — e.g., malformed or oversized image uploads correlated with decoder instability — as a precursor or indicator of exploitation attempts.
Together, these layers let defenders both reduce exposure (patch vulnerable Pillow instances) and detect active exploitation attempts against image-processing services.
Full Detection Details
For the complete detection logic, sample events, investigation guidance, and ready-to-use queries across all seven supported SIEM platforms, see the full detection page: CVE-2026-112403 on df00tech.