CVE-2026-24123: BentoML Path Traversal enables file

Q: Is CVE-2026-24123 actively exploited?

Proof-of-concept exploit code is publicly available for CVE-2026-24123, increasing the risk of exploitation.

Q: How to fix CVE-2026-24123?

1. Patch: Upgrade BentoML to >= 1.4.34 immediately across all build environments. 2. Audit: Inspect bento archives built on affected versions for unexpected non-model files — scan for .ssh, .env, credentials, and config files outside expected directory structures. 3. Trust controls: Never run bentoml build on untrusted bentofiles without prior code review of all file path fields. 4. Detection: Add pre-build linting rules to validate bentofile.yaml path fields against strict allowlists, failing builds on path traversal patterns (e.g., ../). 5. CI/CD hygiene: Ensure build environments are ephemeral, isolated, and credentials are injected via secrets managers rather than stored as on-disk files accessible during builds. 6. Registry audit: Review push history for model registries that received artifacts built by BentoML < 1.4.34 and rotate any credentials that may have been co-located on build hosts.

Q: What systems are affected by CVE-2026-24123?

This vulnerability affects the following AI/ML architecture patterns: MLOps pipelines, model serving, model packaging and distribution, CI/CD build pipelines, model registries.

Q: What is the CVSS score for CVE-2026-24123?

CVE-2026-24123 has a CVSS v3.1 base score of 6.5 (MEDIUM). The EPSS exploitation probability is 0.44%.

CISO Take

If your team builds BentoML packages from external or community-sourced bentofiles, upgrade to 1.4.34 immediately — a malicious bentofile.yaml can silently exfiltrate SSH keys, .env files, and credentials into your bento archive before it reaches a registry. The real risk is supply chain: a compromised open-source model repo can trigger credential theft during a routine CI/CD build. Patch first, then audit any bento archives built from untrusted sources for unexpected file inclusions.

What is the risk?

Rated Medium (CVSS 6.5) but the supply chain angle elevates operational risk for MLOps teams. Confidentiality impact is HIGH — SSH keys and credentials can be silently exfiltrated. EPSS is very low (0.00011), indicating no observed active exploitation. Attack requires user interaction (victim must run bentoml build on a malicious bentofile), limiting blast radius but not eliminating it given the prevalence of community-sourced ML artifacts. Organizations with automated CI/CD pipelines that build third-party BentoML packages face the highest exposure.

What systems are affected?

Package	Ecosystem	Vulnerable Range	Patched
BentoML	pip	—	No patch
8.7K OpenSSF 6.4 21 dependents Pushed 21d ago 55% patched ~14d to patch Full package profile →
BentoML	pip	< 1.4.34	`1.4.34`
8.7K OpenSSF 6.4 21 dependents Pushed 21d ago 55% patched ~14d to patch Full package profile →

How severe is it?

CVSS 3.1

6.5 / 10

EPSS

0.4%

chance of exploitation in 30 days

Higher than 35% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

Exploit Available

Exploitation: MEDIUM

Sophistication

Trivial

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?

AV Network

AC Low

PR None

UI Required

S Unchanged

C High

I None

A None

What should I do?

6 steps

Patch: Upgrade BentoML to >= 1.4.34 immediately across all build environments.
Audit: Inspect bento archives built on affected versions for unexpected non-model files — scan for .ssh, .env, credentials, and config files outside expected directory structures.
Trust controls: Never run bentoml build on untrusted bentofiles without prior code review of all file path fields.
Detection: Add pre-build linting rules to validate bentofile.yaml path fields against strict allowlists, failing builds on path traversal patterns (e.g., ../).
CI/CD hygiene: Ensure build environments are ephemeral, isolated, and credentials are injected via secrets managers rather than stored as on-disk files accessible during builds.
Registry audit: Review push history for model registries that received artifacts built by BentoML < 1.4.34 and rotate any credentials that may have been co-located on build hosts.

What does CISA's SSVC say?

Decision Track

Exploitation none

Automatable No

Technical Impact partial

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

How is it classified?

Supply Chain Data Extraction Framework Inference AML.T0010 - AI Supply Chain Compromise AML.T0010.001 - AI Software AML.T0010.004 - Container Registry AML.T0011 - User Execution AML.T0011.001 - Malicious Package AML.T0025 - Exfiltration via Cyber Means AML.T0037 - Data from Local System AML.T0055 - Unsecured Credentials

Which compliance frameworks are affected?

This CVE is relevant to:

EU AI Act

Art.9 - Risk Management System Article 15 - Accuracy, robustness and cybersecurity Article 9 - Risk management system

ISO 42001

6.1.2 - AI risk assessment 8.4 - AI system security A.6.2 - AI System Lifecycle Processes A.9.4 - Supply Chain Management for AI Systems

NIST AI RMF

GOVERN 1.1 - Organizational Policies for AI Risk GOVERN 6.1 - Policies for AI risk and trustworthiness MANAGE 2.2 - Mechanisms for tracking identified AI risks MANAGE 2.4 - Mechanisms for Tracking Identified AI Risks

OWASP LLM Top 10

LLM05 - Supply Chain Vulnerabilities LLM06 - Sensitive Information Disclosure

Frequently Asked Questions

What is CVE-2026-24123?

If your team builds BentoML packages from external or community-sourced bentofiles, upgrade to 1.4.34 immediately — a malicious bentofile.yaml can silently exfiltrate SSH keys, .env files, and credentials into your bento archive before it reaches a registry. The real risk is supply chain: a compromised open-source model repo can trigger credential theft during a routine CI/CD build. Patch first, then audit any bento archives built from untrusted sources for unexpected file inclusions.

Is CVE-2026-24123 actively exploited?

Proof-of-concept exploit code is publicly available for CVE-2026-24123, increasing the risk of exploitation.

How to fix CVE-2026-24123?

1. Patch: Upgrade BentoML to >= 1.4.34 immediately across all build environments. 2. Audit: Inspect bento archives built on affected versions for unexpected non-model files — scan for .ssh, .env, credentials, and config files outside expected directory structures. 3. Trust controls: Never run bentoml build on untrusted bentofiles without prior code review of all file path fields. 4. Detection: Add pre-build linting rules to validate bentofile.yaml path fields against strict allowlists, failing builds on path traversal patterns (e.g., ../). 5. CI/CD hygiene: Ensure build environments are ephemeral, isolated, and credentials are injected via secrets managers rather than stored as on-disk files accessible during builds. 6. Registry audit: Review push history for model registries that received artifacts built by BentoML < 1.4.34 and rotate any credentials that may have been co-located on build hosts.

What systems are affected by CVE-2026-24123?

This vulnerability affects the following AI/ML architecture patterns: MLOps pipelines, model serving, model packaging and distribution, CI/CD build pipelines, model registries.

What is the CVSS score for CVE-2026-24123?

CVE-2026-24123 has a CVSS v3.1 base score of 6.5 (MEDIUM). The EPSS exploitation probability is 0.44%.

What is the AI security impact?

Affected AI Architectures

MLOps pipelinesmodel servingmodel packaging and distributionCI/CD build pipelinesmodel registries

MITRE ATLAS Techniques

AML.T0010 AI Supply Chain Compromise

AML.T0010.001 AI Software

AML.T0010.004 Container Registry

AML.T0011 User Execution

AML.T0011.001 Malicious Package

AML.T0025 Exfiltration via Cyber Means

AML.T0037 Data from Local System

AML.T0055 Unsecured Credentials

Compliance Controls Affected

EU AI Act: Art.9, Article 15, Article 9

ISO 42001: 6.1.2, 8.4, A.6.2, A.9.4

NIST AI RMF: GOVERN 1.1, GOVERN 6.1, MANAGE 2.2, MANAGE 2.4

OWASP LLM Top 10: LLM05, LLM06

What are the technical details?

Original Advisory

BentoML is a Python library for building online serving systems optimized for AI apps and model inference. Prior to version 1.4.34, BentoML's `bentofile.yaml` configuration allows path traversal attacks through multiple file path fields (`description`, `docker.setup_script`, `docker.dockerfile_template`, `conda.environment_yml`). An attacker can craft a malicious bentofile that, when built by a victim, exfiltrates arbitrary files from the filesystem into the bento archive. This enables supply chain attacks where sensitive files (SSH keys, credentials, environment variables) are silently embedded in bentos and exposed when pushed to registries or deployed. Version 1.4.34 contains a patch for the issue.

Exploitation Scenario

An adversary publishes a popular open-source ML project on GitHub with a crafted bentofile.yaml where the docker.setup_script field contains a path traversal payload pointing to ../../../../home/user/.ssh/id_rsa or a cloud credentials file. A developer or CI/CD pipeline clones the repo and runs bentoml build to package the model for deployment — standard MLOps workflow. BentoML follows the traversal path and silently embeds the SSH private key inside the bento archive without any warning. The team pushes the archive to a shared or public model registry. The adversary pulls the archive and extracts the embedded credential, enabling lateral movement into production infrastructure or cloud environments.

Weaknesses (CWE)

CWE-22 Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal') Primary CWE-22 Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal') Primary

CWE-22 — Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal'): The product uses external input to construct a pathname that is intended to identify a file or directory that is located underneath a restricted parent directory, but the product does not properly neutralize special elements within the pathname that can cause the pathname to resolve to a location that is outside of the restricted directory.

[Implementation] Assume all input is malicious. Use an "accept known good" input validation strategy, i.e., use a list of acceptable inputs that strictly conform to specifications. Reject any input that does not strictly conform to specifications, or transform it into something that does. When performing input validation, consider all potentially relevant properties, including length, type of input, the full range of acceptable values, missing or extra inputs, syntax, consistency across related fields, and conformance to business rules. As an example of business rule logic, "boat" may be syntactically valid because it only contains alphanumeric characters, but it is not valid if the input is only expected to contain colors such as "red" or "blue." Do not rely exclusively on looking for malicious or malformed inputs. This is likely to miss at least one undesirable input, especially if the code's environment changes. This can give attackers enough room to bypass the intended validation. However, denylis
[Architecture and Design] For any security checks that are performed on the client side, ensure that these checks are duplicated on the server side, in order to avoid CWE-602. Attackers can bypass the client-side checks by modifying values after the checks have been performed, or by changing the client to remove the client-side checks entirely. Then, these modified values would be submitted to the server.

Source: MITRE CWE corpus.