CVE-2026-116692: MLflow vulnerable to remote code execution on unauthenticated servers via third-party scorer deserialization importing caller-placed artifact modules
MLflow 3.12.0 up to 3.16.x lets an unauthenticated HTTP caller run Python code on the tracking server. The third-party scorer loader imports a module from a name-prefix allow-list, and the caller can upload a module there through the artifact API. Upgrade to mlflow 3.17.0.
Published Oct 27, 2026 · Analysis Nov 4, 2026
What an attacker can do
From 3.12.0, the server deserializes third-party scorers. It accepts a request-controlled module path under four allowed prefixes (mlflow.genai.scorers.ragas, deepeval, trulens, phoenix), imports it with importlib.import_module, and instantiates a request-controlled class. The check is a string prefix, so any descendant module passes.
The attacker chains built-in endpoints. They create an experiment whose artifact_location points at a writable allowed package directory, create a run, and upload a Python file into the run's artifacts folder. They then call /ajax-api/3.0/mlflow/scorer/invoke with a scorer whose module path resolves to that file. The constructor runs before the server validates trace IDs. The endpoint returns HTTP 500 afterwards, but the code has already run.
On a default no-auth server using the official v3.14.0 image, the researcher got code execution as uid 0. With basic auth enabled, the caller still needs to create experiments and upload artifacts, and the scorer endpoint does no experiment-level permission check. A read-only filesystem, or a package directory the server cannot write, blocks this exact chain. MLflow servers often hold model artifacts, tracking data and cloud credentials, which an attacker would gain.
Who is affected
mlflow from 3.12.0rc0 up to but not including 3.17.0. Version 3.11.1 and earlier lack the feature. The chain needs the server process to write below one of the allowed scorer packages, which is true in the official image and in virtual environments run by the same OS user.
Check your install:
pip show mlflowFor containers, check the image tag, for example ghcr.io/mlflow/mlflow:v3.14.0 is affected.
mlflow
- Ecosystem
- PyPI
- Affected
- >= 3.12.0rc0, < 3.17.0
- Fixed in
- 3.17.0
- Check
pip show mlflow
How to fix it
Upgrade to mlflow 3.17.0:
pip install --upgrade "mlflow>=3.17.0"The fix is in pull request 25906. Until you upgrade, do not expose the tracking server to untrusted networks, turn on authentication, and run the server with a read-only filesystem or with the mlflow package directory owned by a different non-writable user. The advisory says both of these block the exact placement chain.
How to tell if you were hit
Search tracking server logs for POST requests to /ajax-api/3.0/mlflow/scorer/invoke that return HTTP 500, and for experiments created with an artifact_location pointing inside site-packages/mlflow/genai/scorers. Look on disk for run-ID directories under mlflow/genai/scorers/ragas, deepeval, trulens or phoenix containing .py files. Check for unexpected Python files uploaded as run artifacts and for new outbound connections from the server.
Original advisory
From the GitHub Advisory Database (CC BY 4.0).
Summary
MLflow 3.12.0 introduced registration and deserialization support for third-party scorers in OSS MLflow. During scorer deserialization, the server accepts a request-controlled Python module path beneath one of four allowed package prefixes, imports that module, obtains a request-controlled class, and instantiates it synchronously.
An HTTP caller can combine this behavior with MLflow's built-in experiment, run, and artifact-upload endpoints:
- Create an experiment whose local artifact location points to a writable allowed scorer package.
- Create a run, causing MLflow to allocate
<artifact_location>/<run_id>/artifacts/. - Upload a Python module into that run artifact directory.
- Submit a serialized third-party scorer whose module path resolves to the uploaded file.
- Trigger server-side import and class construction through
/ajax-api/3.0/mlflow/scorer/invoke.
On the default no-auth server using the official MLflow v3.14.0 image, this produces unauthenticated arbitrary Python code execution with the privileges of the MLflow server process. The reproduced image runs the process as uid 0 and permits writes to its installed Python package tree.
The attached proof of concept is intentionally harmless. The uploaded constructor performs one action only: it writes a random marker beside its own uploaded run artifact. The marker is then downloaded through MLflow's run-artifact endpoint and checked for an exact match.
Technical details
1. The allow-list accepts attacker-created descendants
At release commit 86cd7f5a1dc25a1387ad07c87811bbc30f62951c, mlflow/genai/scorers/scorer_utils.py defines these allowed roots:
THIRD_PARTY_SCORER_ALLOWED_MODULES = frozenset({
"mlflow.genai.scorers.ragas",
"mlflow.genai.scorers.deepeval",
"mlflow.genai.scorers.trulens",
"mlflow.genai.scorers.phoenix",
})In mlflow/genai/scorers/base.py, the deserializer accepts either an exact root or any descendant:
module_path == m or module_path.startswith(m + ".")It then performs the following operations using request-controlled values:
module = importlib.import_module(module_path)
scorer_class = getattr(module, class_name, None)
scorer_instance = scorer_class(**init_kwargs)A package-name prefix is not a provenance boundary. If the server process can write below an allowed package, an attacker-created descendant can satisfy the string-prefix check and become importable Python code.
Relevant source locations at the v3.14.0 release commit:
mlflow/genai/scorers/scorer_utils.py, allowed third-party scorer modulesmlflow/genai/scorers/base.py, third-party scorer deserialization and construction
2. Deserialization occurs before trace processing
In mlflow/server/handlers.py, _invoke_scorer_handler() parses the supplied serialized scorer and calls:
scorer = Scorer.model_validate_json(serialized_scorer)Only after deserialization and construction does the handler call:
batches = get_trace_batches_for_scorer(trace_ids, scorer, tracking_store)The attacker-controlled constructor therefore runs before the server validates or resolves the supplied trace IDs.
The minimal proof class is deliberately not a complete MLflow scorer. After its constructor writes the marker, later trace-batching logic accesses scorer attributes that the proof class does not implement. The resulting HTTP 500 occurs after code execution and does not invalidate the proof.
No background job execution or special job-execution server option is required to reach the constructor.
3. MLflow's artifact layout places the module under the trusted prefix
The experiment creation endpoint accepts an artifact_location. For file-backed experiments, MLflow derives each run artifact directory as:
<artifact_location>/<run_id>/artifacts/The proof of concept selects this package directory in the official v3.14.0 image:
file:///usr/local/lib/python3.10/site-packages/mlflow/genai/scorers/trulensAfter creating a run and uploading audit_<token>.py, the resulting filesystem location is:
/usr/local/lib/python3.10/site-packages/mlflow/genai/scorers/trulens/<run_id>/artifacts/audit_<token>.pyThe corresponding import path is:
mlflow.genai.scorers.trulens.<run_id>.artifacts.audit_<token>The attack does not overwrite an existing package module. It creates a new run-scoped module in the normal file-backed artifact layout. The run ID is returned to the caller.
4. Authentication and authorization behavior
With authentication disabled, as in the default reproduced server configuration, no credentials are required.
When MLflow basic authentication is enabled, authentication still applies. However, the scorer-invocation AJAX endpoint is assigned to validate_gateway_proxy, which returns True without performing an experiment- or scorer-level resource permission check.
Practical exploitation in an authenticated deployment still requires the caller to create an experiment and run and upload artifacts to them. The key authorization issue is that scorer deserialization itself is not gated by a resource-specific permission check for the supplied experiment before the import occurs.
Proof of concept
The attached reproduction is intentionally isolated:
- exact official MLflow v3.14.0 image pinned by manifest digest;
- Docker network disabled;
- MLflow bound to container loopback;
- no shell command or external callback;
- no external request;
- no host filesystem write;
- marker created beside the uploaded run artifact;
- marker retrieved through the run-artifact download endpoint; and
- disposable container removed after completion.
Run:
./run-official-image.shExpected output:
MLFLOW_VERSION=3.14.0
PACKAGE_DIR=/usr/local/lib/python3.10/site-packages/mlflow/genai/scorers/trulens
CREATE_EXPERIMENT_STATUS=200
CREATE_RUN_STATUS=200
UPLOAD_STATUS=200
INVOKE_STATUS=500
PROOF_DOWNLOAD_STATUS=200
PROOF_MATCH=1
RESULT=AFFECTEDThe security-relevant result is:
PROOF_MATCH=1
RESULT=AFFECTEDThis proves that the attacker-controlled constructor executed in the MLflow server process before the later scorer-processing failure.
Reproduction environment
- Release: MLflow v3.14.0
- Release commit:
86cd7f5a1dc25a1387ad07c87811bbc30f62951c - Official image:
ghcr.io/mlflow/mlflow:v3.14.0 - Manifest digest:
sha256:967de3ca34b6a5751718e380a0a383eb0ce1a6f5b484f363bd0da7559117a5f8 - Python version: 3.10.18
- Image-configured user: empty
- Observed runtime user: uid 0
- Server bind:
127.0.0.1:5000 - Container network: disabled
- Backend store:
sqlite:////work/mlflow.db - Default artifact root:
/work/artifacts - Authentication: disabled
- Special job-execution option: none
Server command inside the disposable container:
mlflow server \
--host 127.0.0.1 \
--port 5000 \
--backend-store-uri sqlite:////work/mlflow.db \
--default-artifact-root /work/artifactsAffected range and source lineage
The vulnerable third-party scorer deserialization behavior was introduced by commit:
02eea26035b9d8e65af31e6e8f058ca195f62cdd
Enable third-party scorer registration in OSS MLflow (#22634)The feature is absent from v3.11.1 and first appears in v3.12.0.
Runtime reproduction succeeds on v3.14.0. The relevant mlflow/genai/scorers/base.py content is unchanged at master commit:
20858c9426a9644dfcf9545a92cbb534376779ebNo fixed release was identified as of 2026-07-12.
Impact
On the reproduced stock no-auth image, any network caller who can reach the tracking server can execute Python code as the MLflow process without credentials or user interaction.
Successful exploitation provides confidentiality, integrity, and availability impact over all data, credentials, network access, and filesystem locations available to the MLflow server process.
The stock image's runtime uid 0 increases container-level impact. Host impact still depends on the container runtime configuration, mounted paths, granted capabilities, socket exposure, and other isolation controls.
In authentication-enabled deployments, exploitation requires an authenticated principal capable of creating an experiment and run and uploading artifacts. The exact chain also requires the server process to be able to write beneath one of the allowed scorer packages. This condition holds in the reproduced official image and may hold in virtual environments installed and run by the same operating-system user.
A read-only container filesystem or a scorer package directory owned by a different, non-writable user blocks this exact placement chain.
Artifacts and attachments
Remote proof-of-concept script
POC Script: new_remote_poc.py
The remote proof of concept was then executed as follows. The disposable machine’s public IP has been redacted:
python3 new_remote_poc.py \
--target http://<REDACTED_PUBLIC_IP> \
--package-dir /usr/local/lib/python3.10/site-packages/mlflow/genai/scorers/trulens \
--acknowledge-owned-target I_OWN_THIS_TARGETThe complete sanitized output was:
TARGET=http://<REDACTED_PUBLIC_IP>
VERSION_STATUS=200
MLFLOW_VERSION=3.14.0
CREATE_EXPERIMENT_STATUS=200
EXPERIMENT_ID=2
CREATE_RUN_STATUS=200
RUN_ID=9a2d665eb38940b68983f0586f623115
UPLOAD_STATUS=200
IMPORT_PATH=mlflow.genai.scorers.trulens.9a2d665eb38940b68983f0586f623115.artifacts.audit_1783843760683972103
INVOKE_STATUS=500
INVOKE_RESPONSE_PREVIEW='<!doctype html>\n<html lang=en>\n<title>500 Internal Server Error</title>\n<h1>Internal Server Error</h1>\nThe server encountered an internal error and was unable to complete your request. Either the server is overloaded or there is an error in the application.\n'
PROOF_DOWNLOAD_STATUS=200
EXPECTED_MARKER_SHA256=7a423b479df3524567fdf8cc3e7f6c4bfccc577fbcc4bac9519db8bd1cc61ed5
PROOF_MATCH=1
SERVER_CLEANUP_PATH=/usr/local/lib/python3.10/site-packages/mlflow/genai/scorers/trulens/9a2d665eb38940b68983f0586f623115
PROOF_OUTPUT_BEGIN
{
"cwd": "/",
"machine": "x86_64",
"marker": "MLFLOW_REMOTE_SAFE_MARKER_1783843760683972103",
"os_release": "PRETTY_NAME=\"Debian GNU/Linux 11 (bullseye)\"\nNAME=\"Debian GNU/Linux\"\nVERSION_ID=\"11\"\nVERSION=\"11 (bullseye)\"\nVERSION_CODENAME=bullseye\nID=debian\nHOME_URL=\"https://www.debian.org/\"\nSUPPORT_URL=\"https://www.debian.org/support\"\nBUG_REPORT_URL=\"https://bugs.debian.org/\"\n",
"platform": "Linux-7.0.0-1006-aws-x86_64-with-glibc2.31",
"python_implementation": "CPython",
"python_version": "3.10.18",
"server_datetime_utc": "2026-07-12T08:09:43.555055+00:00",
"whoami": "root",
"whoami_returncode": 0,
"whoami_stderr": ""
}
PROOF_OUTPUT_END
RESULT=AFFECTED
EXPERIMENT_METADATA_CLEANUP_STATUS=200
CLEANUP_NOTE=discard the temporary test instance or remove SERVER_CLEANUP_PATH on the owned hostThe proof file was also verified directly inside the disposable container:
cat /usr/local/lib/python3.10/site-packages/mlflow/genai/scorers/trulens/9a2d665eb38940b68983f0586f623115/artifacts/execution-proof-1783843760683972103.jsonThe file contained:
{
"cwd": "/",
"machine": "x86_64",
"marker": "MLFLOW_REMOTE_SAFE_MARKER_1783843760683972103",
"os_release": "PRETTY_NAME=\"Debian GNU/Linux 11 (bullseye)\"\nNAME=\"Debian GNU/Linux\"\nVERSION_ID=\"11\"\nVERSION=\"11 (bullseye)\"\nVERSION_CODENAME=bullseye\nID=debian\nHOME_URL=\"https://www.debian.org/\"\nSUPPORT_URL=\"https://www.debian.org/support\"\nBUG_REPORT_URL=\"https://bugs.debian.org/\"\n",
"platform": "Linux-7.0.0-1006-aws-x86_64-with-glibc2.31",
"python_implementation": "CPython",
"python_version": "3.10.18",
"server_datetime_utc": "2026-07-12T08:09:43.555055+00:00",
"whoami": "root",
"whoami_returncode": 0,
"whoami_stderr": ""
}The results demonstrate that:
- the attacker-controlled module was uploaded beneath an allowed scorer package;
- MLflow imported the request-selected module and invoked its constructor;
- the constructor executed as the MLflow server process, which ran as
rootin the reproduced official image; - the constructor created a server-side proof file containing a random marker and bounded runtime information;
- the proof file was retrieved successfully through MLflow’s artifact-download endpoint; and
- the later HTTP 500 occurred only after constructor execution.
The authoritative reproduction indicators are:
UPLOAD_STATUS=200
PROOF_DOWNLOAD_STATUS=200
PROOF_MATCH=1
RESULT=AFFECTEDThe disposable cloud machine and container were removed after testing.
Local Reproduction Scripts: run-official-image.sh poc_local.py
Credited to mskca.
Questions
- Does CVE-2026-116692 need authentication?
Not on a default no-auth server. With basic auth, the caller still needs an account that can create experiments and upload artifacts.
- Is MLflow 3.11.1 affected?
No. The third-party scorer feature first appears in v3.12.0, and the affected range starts at 3.12.0rc0.
- Does a read-only container stop the attack?
It blocks this exact chain, because the module must be written below an allowed scorer package. The advisory says the root cause is still the prefix check.
References
How this page is made: the data panel comes from the GitHub Advisory Database, CISA’s KEV catalog, the NVD and FIRST EPSS, and updates twice a day. The analysis is written by an AI model from the sources listed above and checked by automated rules before it is published. Check versions against the vendor's advisory before you act. Methodology
More MLflow vulnerabilities
All MLflow entriesCVE-2026-116690
MLflow vulnerable to arbitrary file read on unauthenticated servers via prompt model version source
high7.5Oct 27, 2026CVE-2026-116691
MLflow CrewAI autologging records LLM provider API keys in logged trace spans
medium6.5Oct 27, 2026CVE-2026-115886
MLflow: SSRF on unauthenticated MLflow servers via third-party scorer kwargs in the scorer invoke, register, and GenAI evaluate endpoints
medium6.5Oct 23, 2026CVE-2026-53995
MLflow FileStore trace IDs allow path traversal and arbitrary file writes
high8.2Oct 20, 2026GHSA-gqvg-gmmx-x4hm
MLFLOW_ALLOW_PICKLE_DESERIALIZATION=False safety control bypassed by mlflow.statsmodels flavor — RCE via crafted model artifact
high8.8Sep 1, 2026CVE-2026-64849
MLflow: Unauthenticated full-read SSRF in webhook delivery: _validate_webhook_url bypassed via unvalidated HTTP redirects (and DNS rebinding)
critical9.3ExploitedAug 17, 2026
