CVE-2024-37060: MLflow: RCE via deserialization in crafted Recipes

HIGHPoC AVAILABLE

Published June 4, 2024

CISO Take

MLflow Recipes (v1.27.0+) deserialize untrusted data without validation, enabling arbitrary code execution when a data scientist runs a crafted recipe file. Any ML team sharing or consuming MLflow recipes — from public repos, teammates, or vendors — is exposed. Patch MLflow immediately and audit recipe sources; treat untrusted recipe files like executable code.

What is the risk?

CVSS 8.8 with network vector and low complexity makes this high-severity in practice. The user-interaction requirement is a weak barrier in ML workflows where sharing recipe configs is routine and trust is implicit. MLflow's broad adoption across model training, experiment tracking, and CI/CD pipelines dramatically expands the blast radius. No privileges required means any external collaborator or compromised repository can be the delivery vector.

What systems are affected?

Package	Ecosystem	Vulnerable Range	Patched
MLflow	pip	—	No patch
28.6KOpenSSF 5.4695 dependentsPushed 6d ago15% patched~68d to patchFull package profile →

Do you use MLflow? You're affected.

How severe is it?

CVSS 3.1

8.8 / 10

EPSS

0.8%

chance of exploitation in 30 days

Higher than 57% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

Exploit Available

Exploitation: MEDIUM

Sophistication

Moderate

Exploitation Confidence

medium

○Public PoC indexed (trickest/cve)

Composite signal derived from CISA KEV, VulnCheck KEV, CISA SSVC, EPSS, Metasploit, Exploit-DB, trickest/cve, Nuclei templates, and inthewild.io exploitation reports.

What is the attack surface?

AVNetwork

ACLow

PRNone

UIRequired

SUnchanged

CHigh

IHigh

AHigh

What should I do?

6 steps

PATCH
Upgrade MLflow to the latest patched version per the HiddenLayer advisory (https://hiddenlayer.com/sai-security-advisory/mlflow-june2024).
AUDIT
Inventory all recipe files in use — treat them as untrusted code if sourced externally.
RESTRICT
Disable MLflow Recipes feature if not actively used via configuration controls.
ISOLATE
Run MLflow pipelines in sandboxed environments (containers with no credential access, network egress controls).
DETECT
Monitor for unexpected process spawning from MLflow processes; alert on outbound connections from pipeline runners.
SUPPLY CHAIN
Lock recipe sources to internal, version-controlled repositories with code review gates.

What does CISA's SSVC say?

DecisionTrack

Exploitationnone

AutomatableNo

Technical Impacttotal

Source: CISA Vulnrichment (SSVC v2.0). Decision based on the CISA Coordinator decision tree.

How is it classified?

Code ExecutionSupply ChainFrameworkTraining DataAML.T0010.001- AI Software AML.T0011- User Execution AML.T0011.000- Unsafe AI Artifacts AML.T0018.002- Embed Malware AML.T0049- Exploit Public-Facing Application

Which compliance frameworks are affected?

This CVE is relevant to:

EU AI Act

Article 15-Accuracy, robustness and cybersecurityArticle 9-Risk management system

ISO 42001

6.1.2-AI risk assessment8.2-AI system risk management

NIST AI RMF

GOVERN 1.2-Accountability for AI riskMANAGE 2.4-Residual risks are managed

OWASP LLM Top 10

LLM05-Supply Chain Vulnerabilities

Frequently Asked Questions

What is CVE-2024-37060?

Is CVE-2024-37060 actively exploited?

Proof-of-concept exploit code is publicly available for CVE-2024-37060, increasing the risk of exploitation.

How to fix CVE-2024-37060?

1. PATCH: Upgrade MLflow to the latest patched version per the HiddenLayer advisory (https://hiddenlayer.com/sai-security-advisory/mlflow-june2024). 2. AUDIT: Inventory all recipe files in use — treat them as untrusted code if sourced externally. 3. RESTRICT: Disable MLflow Recipes feature if not actively used via configuration controls. 4. ISOLATE: Run MLflow pipelines in sandboxed environments (containers with no credential access, network egress controls). 5. DETECT: Monitor for unexpected process spawning from MLflow processes; alert on outbound connections from pipeline runners. 6. SUPPLY CHAIN: Lock recipe sources to internal, version-controlled repositories with code review gates.

What systems are affected by CVE-2024-37060?

This vulnerability affects the following AI/ML architecture patterns: training pipelines, mlops platforms, model development environments, CI/CD for ML, experiment tracking infrastructure.

What is the CVSS score for CVE-2024-37060?

CVE-2024-37060 has a CVSS v3.1 base score of 8.8 (HIGH). The EPSS exploitation probability is 0.80%.

What is the AI security impact?

Affected AI Architectures

training pipelinesmlops platformsmodel development environmentsCI/CD for MLexperiment tracking infrastructure

MITRE ATLAS Techniques

AML.T0010.001AI Software

AML.T0011User Execution

AML.T0011.000Unsafe AI Artifacts

AML.T0018.002Embed Malware

AML.T0049Exploit Public-Facing Application

Compliance Controls Affected

EU AI Act: Article 15, Article 9

ISO 42001: 6.1.2, 8.2

NIST AI RMF: GOVERN 1.2, MANAGE 2.4

OWASP LLM Top 10: LLM05

What are the technical details?

Original Advisory

Deserialization of untrusted data can occur in versions of the MLflow platform running version 1.27.0 or newer, enabling a maliciously crafted Recipe to execute arbitrary code on an end user’s system when run.

Exploitation Scenario

An adversary publishes a malicious MLflow recipe template to a public GitHub repository or compromises a popular template repo. The recipe YAML embeds a serialized Python object containing a reverse shell payload. A data scientist discovers the template, clones it, and executes `mlflow recipes run` as part of standard workflow. During recipe loading, MLflow deserializes the untrusted object, triggering execution of the embedded payload — establishing persistence, exfiltrating AWS/GCP credentials from environment variables, and potentially poisoning local model artifacts before the user notices anything unusual.

Weaknesses (CWE)

CWE-502Deserialization of Untrusted DataPrimary CWE-502Deserialization of Untrusted Data

CWE-502 — Deserialization of Untrusted Data: The product deserializes untrusted data without sufficiently ensuring that the resulting data will be valid.

[Architecture and Design, Implementation] If available, use the signing/sealing features of the programming language to assure that deserialized data has not been tainted. For example, a hash-based message authentication code (HMAC) could be used to ensure that data has not been modified.
[Implementation] When deserializing data, populate a new object rather than just deserializing. The result is that the data flows through safe input validation and that the functions are safe.

Source: MITRE CWE corpus.