CVE-2026-107736

SumatraPDF is a multi-format reader for Windows. In 3.6.1 and earlier, MaybeFlipBitmap asks GDI+ to read the attacker-controlled PropertyTagOrientation size through GetPropertyItem while supplying only a fixed 64-byte buf on the stack. Opening a crafted TIFF with enough EXIF Orientation values ca...

CVE-2026-107736

SumatraPDF (version 3.6.1 and earlier) on Windows is vulnerable to a stack buffer overflow during the processing of EXIF Orientation metadata. The issue resides in the MaybeFlipBitmap() function, which uses GDI+ to read an attacker-controlled PropertyTagOrientation size via GetPropertyItem() into...

CVE-2026-107736 SumatraPDF: stack buffer overflow while processing EXIF Orientation metadata

SumatraPDF is a multi-format reader for Windows. In 3.6.1 and earlier, MaybeFlipBitmap asks GDI+ to read the attacker-controlled PropertyTagOrientation size through GetPropertyItem while supplying only a fixed 64-byte buf on the stack. Opening a crafted TIFF with enough EXIF Orientation values ca...

PT-2026-108304

Name of the Vulnerable Software and Affected Versions SumatraPDF versions prior to 3.6.2 Description The MaybeFlipBitmap function uses GetPropertyItem to read the PropertyTagOrientation size via GDI+ into a fixed 64-byte buffer on the stack. An attacker can trigger a stack-based buffer overflow b...

Open Babel has out-of-bounds write in Gaussian coords_type orientation parser

SummaryA memory-safety vulnerability in Open Babel's Gaussian output parserallowed an out-of-bounds write when reading a crafted input file. DetailsThe flaw was in the coordstype orientation parser inside theGaussian output reader. A malformed orientation block caused theparser to write past the...

PYSEC-2026-3406 vLLM: image EXIF Rotation & PNG tRNS Transparency Not Normalized, Causing Mismatch Between Model Input and Expectations

Summary Issue 1: EXIF orientation not normalized → The image orientation processed by the model differs from how humans view it, introducing interpretation bias. Issue 2: PNG tRNS not explicitly flattened before converting to RGB → After conversion, transparent/semi-transparent pixels are rendere...

vLLM: image EXIF Rotation & PNG tRNS Transparency Not Normalized, Causing Mismatch Between Model Input and Expectations

SummaryIssue 1: EXIF orientation not normalized → The image orientation processed by the model differs from how humans view it, introducing interpretation bias.Issue 2: PNG tRNS not explicitly flattened before converting to RGB → After conversion, transparent/semi-transparent pixels are rendered...

PT-2026-60018

Summary Issue 1: EXIF orientation not normalized → The image orientation processed by the model differs from how humans view it, introducing interpretation bias. Issue 2: PNG tRNS not explicitly flattened before converting to RGB → After conversion, transparent/semi-transparent pixels are rendere...

arucophotoembedding (0.0.1), arvideo (0.0.1) +4 more potentially affected by CVE-2019-5063 via opencv-contrib-python (>=4.0.0.21, <=4.1.0.25)

opencv-contrib-python PYPI version =4.0.0.21, =0.0.1, =0.0.2, =0.9.1, =0.0.12, =0.0.13 Source cves: CVE-2019-5063 Source advisory: OSV:PYSEC-2026-2801...

arucophotoembedding (0.0.1), arvideo (0.0.1) +4 more potentially affected by CVE-2019-15939 via opencv-contrib-python (>=4.0.0.21, <=4.1.0.25)

opencv-contrib-python PYPI version =4.0.0.21, =0.0.1, =0.0.2, =0.9.1, =0.0.12, =0.0.13 Source cves: CVE-2019-15939 Source advisory: OSV:PYSEC-2026-2799...

arucophotoembedding (0.0.1), arvideo (0.0.1) +4 more potentially affected by CVE-2019-14492 via opencv-contrib-python (>=4.0.0.21, <=4.1.0.25)

opencv-contrib-python PYPI version =4.0.0.21, =0.0.1, =0.0.2, =0.9.1, =0.0.12, =0.0.13 Source cves: CVE-2019-14492 Source advisory: OSV:PYSEC-2026-2798...

arucophotoembedding (0.0.1), arvideo (0.0.1) +4 more potentially affected by CVE-2019-14493 via opencv-contrib-python (>=4.0.0.21, <=4.1.0.25)

opencv-contrib-python PYPI version =4.0.0.21, =0.0.1, =0.0.2, =0.9.1, =0.0.12, =0.0.13 Source cves: CVE-2019-14493 Source advisory: OSV:PYSEC-2026-2795...

Open Babel has out-of-bounds write in Gaussian coords_type orientation parser

Summary A memory-safety vulnerability in Open Babel's Gaussian output parser allowed an out-of-bounds write when reading a crafted input file. Details The flaw was in the coordstype orientation parser inside the Gaussian output reader. A malformed orientation block caused the parser to write past...

PT-2026-55163

Summary A memory-safety vulnerability in Open Babel's Gaussian output parser allowed an out-of-bounds write when reading a crafted input file. Details The flaw was in the coords type orientation parser inside the Gaussian output reader. A malformed orientation block caused the parser to write pas...

GHSA-X8XR-MJ9X-6H7W Duplicate Advisory: image EXIF Rotation & PNG tRNS Transparency Not Normalized, Causing Mismatch Between Model Input and Expectations

Duplicate Advisory This advisory has been withdrawn because it is a duplicate of GHSA-8jr5-v98p-w75m. This link is maintained to preserve external references. Original Description A flaw was found in vLLM, an open-source library for large language model inference. This vulnerability arises from...

EUVD-2026-37645

A flaw was found in vLLM, an open-source library for large language model inference. This vulnerability arises from improper handling of image metadata, specifically EXIF orientation and PNG transparency tRNS data, during image processing. When images are converted to RGB, transparency informatio...

Duplicate Advisory: image EXIF Rotation & PNG tRNS Transparency Not Normalized, Causing Mismatch Between Model Input and Expectations

Duplicate Advisory This advisory has been withdrawn because it is a duplicate of GHSA-8jr5-v98p-w75m. This link is maintained to preserve external references. Original Description A flaw was found in vLLM, an open-source library for large language model inference. This vulnerability arises from...

GHSA-8JR5-V98P-W75M vLLM: image EXIF Rotation & PNG tRNS Transparency Not Normalized, Causing Mismatch Between Model Input and Expectations

Summary Issue 1: EXIF orientation not normalized → The image orientation processed by the model differs from how humans view it, introducing interpretation bias. Issue 2: PNG tRNS not explicitly flattened before converting to RGB → After conversion, transparent/semi-transparent pixels are rendere...

Interpretation Conflict

Overview vllm is an A high-throughput and memory-efficient inference and serving engine for LLMs Affected versions of this package are vulnerable to Interpretation Conflict in the image processing pipeline. An attacker can cause the model to interpret images differently from human expectations by...

CVE-2026-12491

A flaw was found in vLLM, an open-source library for large language model inference. This vulnerability arises from improper handling of image metadata, specifically EXIF orientation and PNG transparency tRNS data, during image processing. When images are converted to RGB, transparency informatio...