dbcve.org
Megatron LmApplication · Nvidia

CVE-2025-23348

HIGH · 7.8 CVSS v3.1 Published 2025-09-24
Fix available
A fix is available. Upgrade to 0.12.3 or later.
See remediation →
80/100
Remediation priority · High
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 Megatron-LM for all platforms contains a vulnerability in the pretrain_gpt script, where malicious data created by an attacker may cause a code injection issue. A successful exploit of this vulnerability may lead to code execution, escalation of privileges, information disclosure, and data tampering.

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 Megatron-LM contains a code injection vulnerability in the pretrain_gpt training script. The vulnerability allows an attacker to craft malicious training data that gets improperly interpreted and executed, leading to arbitrary code execution on the target system. This is a classic data injection attack vector where untrusted input is not properly sanitized before being processed or evaluated.

MitigationImplement strict input validation and sanitization on all training data before processing. Use safe parsing methods (e.g., ast.literal_eval instead of eval, avoid os.system/popen with user data). Apply principle of least privilege to the training process and consider sandboxing data loading operations.

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
Megatron LmApplication
Affected:< 0.12.3= 0.13.0

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
High
Availability
High

CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/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 Megatron-LM version
    Run 'pip show megatron-lm' or check the version in your environment (e.g., via Python: import megatron; print(megatron.__version__))
    Affected if The version is lower than 0.12.3 or exactly 0.13.0
  2. Confirm pretrain_gpt script is in use
    Check if your training pipeline invokes the pretrain_gpt.py script or a script that imports and uses the GPT pretraining functionality from Megatron-LM
    Affected if You are running or importing the pretrain_gpt training component in your pipeline
  3. Verify training data handling
    Inspect how training data is loaded and processed in your pretrain_gpt execution context - examine any custom data loading code or configurations that feed data to the model
    Affected if Training data (especially from untrusted or external sources) is being processed by the pretrain_gpt script in an affected version

You are affected if you are running a vulnerable Megatron-LM version (below 0.12.3 or exactly 0.13.0) and using the pretrain_gpt training script with training data that could be externally manipulated.

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 0.12.3 or later
Fixed in 0.12.3
Interim mitigation

Implement strict input validation and sanitization on all training data before processing. Use safe parsing methods (e.g., ast.literal_eval instead of eval, avoid os.system/popen with user data). Apply principle of least privilege to the training process and consider sandboxing data loading operations.

Recommended fix Moderate confidence

Upgrade to Megatron-LM version 0.14.0 or later (or the latest stable release)

  1. 1. Identify the current installed version of Megatron-LM by checking the version file or running pip show megatron-lm
  2. 2. If running from source, check the git tag or version file in the repository
  3. 3. Ensure any dependent packages are compatible with the target version before upgrading
  4. 4. Backup any custom scripts, configurations, or data that interact with the pretrain_gpt script
  5. 5. Upgrade Megatron-LM to version 0.14.0 or later using: pip install megatron-lm>=0.14.0 (or pip install --upgrade megatron-lm)
  6. 6. Verify the upgrade was successful by checking the installed version
  7. 7. Test the pretrain_gpt script functionality with your pipeline to ensure the code injection vulnerability is remediated
Caveat Review NVIDIA's release notes for 0.14.0 for any API changes or migration requirements from earlier versions

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

Fix this in Megatron Lm Scoped from the published advisory
  • Consultation3.0 h
  • Implementation4.0 h
  • Testing2.0 h
  • Review / QA2.0 h
11.0 hours of engineering $1,980
Get the upgrade done

An estimate, not a bill — we confirm scope with you before any work starts. Need it this week? Rush from $3,168.

Scan for this in your stack

Free · runs locally
dbcve dependency scanner

Check whether your project pulls in CVE-2025-23348 — 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-2025-23348 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