dbcve.org
Cuda ToolkitApplication · Nvidia

CVE-2025-23275

HIGH · 7.1 CVSS v3.1 Published 2025-09-24
Fix available
A fix is available. Upgrade to 13.0.0 or later.
See remediation →
73/100
Remediation priority · Elevated
Zero-click

Official description Straight from the sourceThe vendor's or NVD's own wording, published unedited. Authoritative, but often terse — it says what broke, rarely what to do.

NVD · unedited
NVIDIA CUDA Toolkit for all platforms contains a vulnerability in nvJPEG where a local authenticated user may cause a GPU out-of-bounds write by providing certain image dimensions. A successful exploit of this vulnerability may lead to denial of service and information disclosure.

Technical summary Written by usOur analysis, written from the advisory, the CVSS vector and the affected-version data. It adds context the advisory leaves out, and never invents facts that are not in the source.

dbcve analysis · moderate confidence

NVIDIA CUDA Toolkit's nvJPEG library contains an out-of-bounds write vulnerability where insufficient validation of image dimensions allows a local authenticated user to trigger memory corruption on the GPU. By providing specially crafted image dimension values, an attacker can write beyond allocated buffer boundaries, potentially causing denial of service through GPU instability or enabling information disclosure via memory exposure.

MitigationApply the NVIDIA CUDA Toolkit security patch when released. Until then, restrict local user access to systems with CUDA installations and implement input validation on image dimensions before passing to nvJPEG functions.

Verify against the referenced sources before acting — the references below are authoritative for this CVE, this summary is not.

Affected products & versions What the vendor confirmedThe version ranges the vendor confirmed as vulnerable. If your version sits inside a range here, treat yourself as exposed until you have upgraded.

NVD · CPE data
Cuda ToolkitApplication
Affected:< 13.0.0
NvjpegApplication
Affected:all versions

CVSS breakdown How the score is builtThe industry scoring standard. It rates how the flaw is reached, what it takes to exploit, and what an attacker gains — the score is derived from those, not the other way round.

From the vector
Attack vector
Local
Complexity
Low
Privileges
Low
User interaction
None
Scope
Unchanged
Confidentiality
High
Integrity
None
Availability
High

CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:H

Am I affected? How to checkSteps we derive from the advisory and the affected-version data, so you can decide whether this CVE reaches your setup. They are a guide, not a scan — your own configuration is the authority.

dbcve checks

Work through these to decide whether this CVE applies to you.

  1. Identify installed CUDA Toolkit version
    Run 'nvcc --version' or check NVIDIA GPU driver info, or look for CUDA installation directories (e.g., /usr/local/cuda, C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA)
    Affected if CUDA Toolkit version is less than 13.0.0
  2. Locate nvJPEG library files
    Search for nvjpeg library files: on Linux 'find /usr -name "*nvjpeg*" 2>/dev/null', on Windows check CUDA bin/lib directories for nvjpeg.dll or nvjpeg.lib
    Affected if nvjpeg library files exist on the system (all versions are affected)
  3. Identify applications using nvJPEG
    Scan installed software for binaries that link to nvjpeg library using 'ldd' on Linux or Dependency Walker on Windows, or check application documentation for CUDA/nvJPEG usage
    Affected if Applications link against or call nvJPEG functions (nvjpegDecode, nvjpegEncode, etc.)
  4. Check if image dimension validation is performed before nvJPEG calls
    Review source code or configuration of applications using nvJPEG to determine if they validate image width/height parameters before passing to nvJPEG decode/encode functions
    Affected if Applications pass image dimensions to nvJPEG without validating that dimensions are within safe bounds
  5. Verify local user access to CUDA-enabled systems
    Audit system permissions to determine if untrusted local users have access to systems with CUDA installations
    Affected if Untrusted local users can execute code on systems with vulnerable nvJPEG library present

A system is affected if CUDA Toolkit version is below 13.0.0 AND any application uses nvJPEG library to process images without validating image dimensions beforehand.

Generated from the published advisory. Verify against your own configuration.

Check your environment

Paste your version and any relevant configuration and it will be compared against the affected criteria above. Do not include secrets or credentials.

AI-assisted, checked against the advisory. Informational, not a guarantee.

Remediation Closing itWhat it takes to close this. Where a vendor fix exists we point at it; where none exists we say so plainly, and can build one. Effort estimates are scoped from the advisory, not from your codebase.

dbcve · scoped
Upgrade available Upgrade to 13.0.0 or later
Fixed in 13.0.0
Interim mitigation

Apply the NVIDIA CUDA Toolkit security patch when released. Until then, restrict local user access to systems with CUDA installations and implement input validation on image dimensions before passing to nvJPEG functions.

Recommended fix High confidence

CUDA Toolkit 13.0.0 or later

  1. 1. Check current CUDA Toolkit version by running 'nvcc --version' or 'nvidia-smi'
  2. 2. Download CUDA Toolkit 13.0.0 or later from the official NVIDIA website (developer.nvidia.com/cuda-downloads)
  3. 3. Uninstall the current CUDA Toolkit version if installed via packages
  4. 4. Install the new CUDA Toolkit 13.0.0+ using the installer for your platform
  5. 5. Update PATH and LD_LIBRARY_PATH environment variables to point to the new installation if necessary
  6. 6. Verify installation by running 'nvcc --version' and confirming it shows version 13.0.0 or later
  7. 7. Rebuild and test any applications that use nvJPEG to ensure compatibility
Caveat CUDA 13.0 is a major version upgrade with potential API changes, driver compatibility updates, and deprecated features that may require code modifications

Generated from the published advisory — verify against the referenced sources before acting.

Fix this in Cuda Toolkit Scoped from the published advisory
  • Consultation4.0 h
  • Implementation2.0 h
  • Testing8.0 h
  • Review / QA3.0 h
17.0 hours of engineering $2,900
Get the upgrade done

An estimate, not a bill — we confirm scope with you before any work starts. Need it this week? Rush from $4,640.

Scan for this in your stack

Free · runs locally
dbcve dependency scanner

Check whether your project pulls in CVE-2025-23275 — or any other known-vulnerable package — straight from your lock files. Free and open source; it runs locally and uploads nothing.

Get it on GitHub Scan in the browser

References Go to the primary sourcePrimary sources — vendor advisories, patches and trackers. Where our summary and a reference disagree, the reference wins.

Primary sources

Practitioner notes

Contributed

Peer-ranked notes from engineers who’ve handled CVE-2025-23275 in production — separate from our analysis above.

No notes yet

Be the first to add a field note for this CVE — a mitigation you’ve verified, a version caveat, or a link to a working fix. Sign in above to contribute.

What this is

A place for practitioners to share what actually worked: a mitigation you’ve tested, a configuration change, a version- or environment-specific caveat, or a link to a verified patch. The most useful notes rise to the top as peers upvote them, so the signal stays high.

What belongs here
  • Verified mitigations, workarounds, and config changes
  • Version or environment caveats, and links to real fixes
  • No weaponised exploit code, or anything meant to cause harm
  • No spam, self-promotion, credentials, or personal data