dbcve.org
Code InjectionWeakness · CWE-94

CVE-2026-46432

HIGH · 7.8 CVSS v3.1 Published 2026-06-10
Mitigation only
No fix yet — a mitigation exists. There is no fixed release. A documented workaround reduces exposure in the meantime.
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
LMDeploy is a toolkit for compressing, deploying, and serving large language models. In versions 0.12.3 and prior, LMDeploy is vulnerable to arbitrary code execution through hardcoded "trust_remote_code=True" in multiple HuggingFace model-loading call sites. At time of publication, there are no publicly available patches.

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 · high confidence

LMDeploy versions 0.12.3 and prior contain hardcoded 'trust_remote_code=True' in HuggingFace model-loading call sites. This setting allows arbitrary code execution if a user loads a malicious model, as the model's code executes within LMDeploy's context without validation.

MitigationUntil an official patch is released, avoid loading models from untrusted sources. Manually review and modify the codebase to disable trust_remote_code=True or implement model validation before loading.

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

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. Check installed LMDeploy version
    Run 'pip show lmdeploy' or execute 'python -c "import lmdeploy; print(lmdeploy.__version__)"' to retrieve the installed version
    Affected if The installed version is 0.12.3 or lower
  2. Locate HuggingFace model-loading call sites
    Search the LMDeploy codebase for model loading functions such as from_pretrained, pipeline, or AutoModel calls that interact with HuggingFace models
    Affected if Any HuggingFace model loading call sites are found in the codebase
  3. Inspect trust_remote_code parameter in model loading calls
    Examine each HuggingFace model-loading call site to determine whether trust_remote_code is explicitly set to True or defaults to True
    Affected if Any model-loading call site hardcodes or defaults trust_remote_code=True

You are affected if LMDeploy version is 0.12.3 or prior AND any HuggingFace model loading call sites in your deployment use trust_remote_code=True to load models.

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
Mitigation available No clean upgrade yet — mitigate in the meantime
Mitigation

Until an official patch is released, avoid loading models from untrusted sources. Manually review and modify the codebase to disable trust_remote_code=True or implement model validation before loading.

Recommended fix High confidence
  1. 1. Audit all codebases using LMDeploy for HuggingFace model loading calls (model.from_pretrained or similar)
  2. 2. Identify all instances where trust_remote_code is not explicitly set (defaults to True) or explicitly set to True
  3. 3. For each model loading call, explicitly set trust_remote_code=False if the model does not require custom code execution
  4. 4. If a model legitimately requires trust_remote_code=True for custom model code, verify the model's provenance: ensure it is from a trusted source, review the model's code for malicious content, and consider hosting a local copy
  5. 5. Implement a global configuration in LMDeploy to default trust_remote_code to False across all model loading operations
  6. 6. Add model source allow-listing/verification to ensure only trusted HuggingFace repos or local paths are loaded
  7. 7. Implement logging/monitoring around model loading to detect unauthorized code execution attempts
Caveat Setting trust_remote_code=False may break models that legitimately require custom code execution - such models will need to be reviewed and either replaced with safe alternatives or hosted locally with verified code

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

Have this fixed Scoped from the published advisory
  • Consultation3.0 h
  • Implementation12.0 h
  • Testing6.0 h
  • Review / QA4.0 h
25.0 hours of engineering $4,380
Get help mitigating

An estimate, not a bill — we confirm scope with you before any work starts. Need it this week? Rush from $7,008.

Scan for this in your stack

Free · runs locally
dbcve dependency scanner

Check whether your project pulls in CVE-2026-46432 — 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-2026-46432 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