CVE-2026-116692
MLflow: Remote code execution on unauthenticated MLflow servers via third-party scorer deserialization importing caller-placed artifact modules
No known exploitation. EPSS puts it in the 53rd percentile. No fix published yet.
No public exploit knownFixNone yetNo fix or workaround published
No ready-made check. Neither Nuclei nor OpenVAS has one for this CVE yet.
What to do
No fixed build or workaround is published yet. Limit exposure and watch for a patch.
- MlflowAffected: >= 3.12.0rc0, < 3.17.0Upgrade to: not statedStated by: CVE record
Versions as the sources state them. Full version matrix after sign-in
Steps
Written by AI from the record- Check whether you run MLflow, and whether it is reachable from the internet or an untrusted network.
- If authentication is disabled on your MLflow server, treat this as high risk and restrict network access immediately (firewall, VPN-only, or take it offline until secured).
- If basic authentication is enabled, review who has permission to create experiments, runs, and upload artifacts — limit this to trusted users only.
- There is currently no confirmed patched version from the vendor in our records, so ask your IT team or MLflow vendor directly for the fixed release and apply it as soon as it's available.
- In the meantime, disable or restrict the scorer invoke feature if your deployment doesn't need it.
What it is
From the CVE record
MLflow is an open source AI engineering platform for agents, large language models, and machine learning models. From 3.12.0rc0 until 3.17.0, third-party scorer deserialization in mlflow/genai/scorers/base.py accepts request-selected descendants of THIRD_PARTY_SCORER_ALLOWED_MODULES rather than requiring an exact allowed module. A caller can use MLflow artifact endpoints to place a Python module beneath a writable allowed package and submit its module and class through /ajax-api/3.0/mlflow/scorer/invoke. Scorer.model_validate_json() imports the module and constructs the class before trace processing. Authentication-disabled deployments permit this code execution without credentials, while basic-auth deployments require a principal capable of creating an experiment and run and uploading artifacts. In either configuration, the allowed scorer package must be writable by the server process. This issue is fixed in version 3.17.0.
In plain language
Written by AI from the recordThis flaw lets an attacker take over an MLflow server used to manage AI/ML models, but it isn't being actively exploited and there's no confirmed fix yet, so it needs attention soon rather than emergency action tonight.
Unauthenticated remote code execution in MLflow (CVE-2026-116692) arises because the third-party scorer deserialization path in mlflow/genai/scorers/base.py accepts any descendant of an allowed module list instead of an exact match, letting a caller upload a malicious module via artifact endpoints and trigger its import/instantiation through the scorer invoke API.
If you're affected
- Full server takeover
- Exposure of ML models and data
- Pivot point into internal network
- Operational disruption of AI workflows
Exploitation
Where each signal puts this CVE on the scale from published to confirmed exploited.
- CISA KEV
Not in the catalog. CISA has not confirmed exploitation.
- Public exploits
No public exploit or proof of concept found in the sources we track.
- EPSS
0.7% chance of exploitation activity in the next 30 days, which ranks it in the 53rd percentile of scored CVEs.
Exploit Prediction Scoring System, FIRST.org. A probability, not a confirmation.
- Attention now
Sustained.
Lifecycle
6 events over 6 days, from the signal feeds we watch.
- Record updated
- Record updated
- Publishedweakness classified, att&ck mapped, record updated
Affected products
Technical detail
CVSS 3.1 vector
Open in the CVSS calculatorCVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
Scored 9.8 by NVD.
How it is reached
- Attack Vector NetworkExploitable remotely over the network without any special conditions
- Attack Complexity LowNo special conditions — the attack can be reliably reproduced
- Privileges Required NoneNo authentication required — anyone can exploit this
- User Interaction NoneNo user interaction needed — fully automated exploitation
Scope
- Scope UnchangedImpact is limited to the vulnerable component itself
Impact if exploited
- Confidentiality HighTotal information disclosure — all data in the component is compromised
- Integrity HighTotal loss of integrity — attacker can modify any data in the component
- Availability HighTotal denial of service — the component is completely unavailable
Weaknesses
ATT&CK techniques
Mapped from the weaknesses above (CWE to ATT&CK), not observed in attacks.
Sources
Watch the software you run.
My Stack ranks new CVEs for your products by real-world exploitation, so the next serious one reaches you without reading every advisory.
We'll flag the next CVE, public exploit or patch for Mlflow, not every advisory.
A free account adds
- The full version matrix and every affected product
- Exploit links, proofs of concept and Metasploit modules
- Email alerts for the products you watch
- The same data over REST API, MCP and CLI