Code InjectionWeakness · CWE-94

CVE-2026-47117

CRITICAL · 9.8 CVSS v3.1 Published 2026-06-02
Patch available
A vendor patch is available. No clean upgrade release — apply the published patch.
See remediation →
100/100
Remediation priority · Urgent
Remotely reachable No privileges Zero-click Patch available

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
OpenMed before 1.5.2 contains a remote code execution vulnerability in the PII privacy-filter model loading path. The privacy-filter dispatcher used broad substring matching on the user-supplied model_name parameter, allowing a value such as attacker/foo-privacy-filter-bar to route through a path that loads Hugging Face models with trust_remote_code=True. An unauthenticated attacker can supply a malicious model repository containing custom Transformers code via auto_map in config.json or tokenizer_config.json, which is imported and executed with the privileges of the OpenMed service process.

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

OpenMed before 1.5.2 has a critical RCE vulnerability where the privacy-filter dispatcher uses broad substring matching on user-supplied model_name parameters, allowing values like 'attacker/foo-privacy-filter-bar' to route to HuggingFace model loading with trust_remote_code=True. Unauthenticated attackers can supply malicious model repositories containing executable Transformers code via auto_map in config.json or tokenizer_config.json, which gets imported and executed with the service process privileges.

MitigationUpgrade to OpenMed 1.5.2 or later which fixes the substring matching logic and restricts trust_remote_code. If immediate upgrade is not possible, disable the privacy-filter model loading functionality entirely.

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
Network
Complexity
Low
Privileges
None
User interaction
None
Scope
Unchanged
Confidentiality
High
Integrity
High
Availability
High

CVSS:3.1/AV:N/AC:L/PR:N/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 OpenMed version
    Run 'openmed --version' or check the installed package version via pip/pip3 list | grep openmed, or inspect the version file in the OpenMed installation directory
    Affected if The installed version is before 1.5.2 (e.g., 1.5.1, 1.5.0, or earlier)
  2. Locate privacy-filter model loading code
    Search the codebase for references to 'privacy-filter' or the model loading function that handles Hugging Face model loading (look for keywords like 'model_name', 'from_pretrained', 'trust_remote_code')
    Affected if The privacy-filter model loading feature exists and is accessible in the deployment
  3. Check model_name parameter validation
    Inspect the dispatcher or routing code that processes the model_name parameter. Look for substring matching logic that handles requests containing 'privacy-filter'. Verify if strict allowlist validation is implemented
    Affected if The code uses broad substring matching (e.g., 'if "privacy-filter" in model_name') rather than exact matches or an allowlist
  4. Verify trust_remote_code setting
    Search the model loading code for Hugging Face model loading calls (from_pretrained or similar). Check if trust_remote_code=True is passed, which allows execution of code from loaded model repositories
    Affected if trust_remote_code=True is set when loading models from user-supplied model names
  5. Check network exposure
    Review network configuration and firewall rules to determine if the server running OpenMed can make outbound connections to Hugging Face (huggingface.co) or other model repositories
    Affected if Outbound network access to untrusted Hugging Face repositories is permitted from the OpenMed server

You are affected if OpenMed version is below 1.5.2 AND the privacy-filter model loading feature is in use with trust_remote_code=True enabled and the model_name parameter lacks strict allowlist validation.

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
Patch available Apply the vendor patch
Vendor patch github.com →
Interim mitigation

Upgrade to OpenMed 1.5.2 or later which fixes the substring matching logic and restricts trust_remote_code. If immediate upgrade is not possible, disable the privacy-filter model loading functionality entirely.

Recommended fix Moderate confidence

OpenMed version 1.5.2

  1. 1. Identify the currently deployed OpenMed version by checking the codebase or deployment artifacts
  2. 2. If the current version is before 1.5.2, plan an upgrade to version 1.5.2 or later
  3. 3. Review the vendor patch at https://github.com/maziyarpanahi/openmed/commit/98724f65df98d7518b9006e6356740aa36c2f224 to understand the fix (likely adds proper validation of the model_name parameter to prevent path traversal and restricts trust_remote_code=True usage)
  4. 4. Before upgrading in production, test the new version in a staging environment to verify the vulnerability is remediated
  5. 5. Deploy version 1.5.2 or latest stable release that includes this fix
  6. 6. After upgrading, verify that the privacy-filter functionality still works correctly with legitimate models
  7. 7. Monitor for any unusual behavior or unauthorized code execution attempts

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

Have this fixed Scoped from the published advisory
  • Consultation6.0 h
  • Implementation12.0 h
  • Testing8.0 h
  • Review / QA4.0 h
30.0 hours of engineering $5,280
Get the patch applied

An estimate, not a bill — we confirm scope with you before any work starts. Need it this week? Rush from $8,448.

Scan for this in your stack

Free · runs locally
dbcve dependency scanner

Check whether your project pulls in CVE-2026-47117 — or any other known-vulnerable package — straight from your lock files. Free and open source; it runs locally and uploads nothing.

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-2026-47117 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