CVE-2022-41891: TensorFlow segfault DoS

Q: Is CVE-2022-41891 actively exploited?

Proof-of-concept exploit code is publicly available for CVE-2022-41891, increasing the risk of exploitation.

Q: How to fix CVE-2022-41891?

1. Patch: Upgrade to TensorFlow 2.11, 2.10.1, 2.9.3, or 2.8.4 — cherry-pick commit fc33f3dc4c14051a83eec6535b608abe1d355fde if running a custom build. 2. Short-term workaround: Add input validation middleware to inference endpoints that rejects requests with empty element_shape tensors before they reach the TF runtime. 3. Network hardening: Ensure TF Serving is not directly internet-exposed; place behind an API gateway or load balancer that performs basic schema validation. 4. Detection: Monitor for abnormal process restarts or segfault signals in TF Serving logs; alert on SIGSEGV in inference worker processes. 5. Verify: Run `python -c "import tensorflow as tf; print(tf.__version__)"` across all inference nodes to confirm patched versions.

Q: What systems are affected by CVE-2022-41891?

This vulnerability affects the following AI/ML architecture patterns: model serving, training pipelines, inference, batch prediction pipelines.

Q: What is the CVSS score for CVE-2022-41891?

CVE-2022-41891 has a CVSS v3.1 base score of 7.5 (HIGH). The EPSS exploitation probability is 0.43%.

CISO Take

Any TensorFlow-based model serving endpoint is remotely crashable with a single malformed request — no authentication required. If your inference infrastructure runs TF < 2.11/2.10.1/2.9.3/2.8.4 and is network-accessible, patch immediately. This is a trivial availability kill-switch for production ML pipelines.

What is the risk?

High risk for organizations with network-exposed TensorFlow serving endpoints. CVSS 7.5 with AV:N/AC:L/PR:N/UI:N means unauthenticated remote exploitation with near-zero complexity — essentially a one-liner crash trigger. The blast radius is limited to availability (no C/I impact), but in production AI inference contexts, a DoS against model serving equates to direct business disruption. Risk is elevated because TF Serving is commonly exposed internally across large ML platforms without input validation layers.

What systems are affected?

Package	Ecosystem	Vulnerable Range	Patched
TensorFlow	pip	—	No patch
195.8K OpenSSF 7.1 3.7K dependents Pushed 3d ago 4% patched ~1372d to patch Full package profile →

Do you use TensorFlow? You're affected.

How severe is it?

CVSS 3.1

7.5 / 10

EPSS

0.4%

chance of exploitation in 30 days

Higher than 34% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

Exploit Available

Exploitation: MEDIUM

Sophistication

Trivial

Exploitation Confidence

medium

○ CISA SSVC: Public PoC

○ 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 None

S Unchanged

C None

I None

A High

What should I do?

5 steps

Patch: Upgrade to TensorFlow 2.11, 2.10.1, 2.9.3, or 2.8.4 — cherry-pick commit fc33f3dc4c14051a83eec6535b608abe1d355fde if running a custom build.
Short-term workaround: Add input validation middleware to inference endpoints that rejects requests with empty element_shape tensors before they reach the TF runtime.
Network hardening: Ensure TF Serving is not directly internet-exposed; place behind an API gateway or load balancer that performs basic schema validation.
Detection: Monitor for abnormal process restarts or segfault signals in TF Serving logs; alert on SIGSEGV in inference worker processes.
Verify: Run python -c "import tensorflow as tf; print(tf.__version__)" across all inference nodes to confirm patched versions.

What does CISA's SSVC say?

Decision Track*

Exploitation poc

Automatable No

Technical Impact partial

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

How is it classified?

DoS Framework Inference AML.T0010.001 - AI Software AML.T0029 - Denial of AI Service AML.T0049 - Exploit Public-Facing Application

Which compliance frameworks are affected?

This CVE is relevant to:

EU AI Act

Art.15 - Accuracy, robustness and cybersecurity for high-risk AI systems

ISO 42001

A.6.2.6 - AI system security and resilience

NIST AI RMF

GOVERN-1.7 - Processes for tracking identified AI risks MANAGE-2.4 - Mechanisms for vulnerability remediation and incident response

OWASP LLM Top 10

LLM09:2025 - Misinformation / Overreliance on AI output (via service disruption)

Frequently Asked Questions

What is CVE-2022-41891?

Any TensorFlow-based model serving endpoint is remotely crashable with a single malformed request — no authentication required. If your inference infrastructure runs TF < 2.11/2.10.1/2.9.3/2.8.4 and is network-accessible, patch immediately. This is a trivial availability kill-switch for production ML pipelines.

Is CVE-2022-41891 actively exploited?

Proof-of-concept exploit code is publicly available for CVE-2022-41891, increasing the risk of exploitation.

How to fix CVE-2022-41891?

1. Patch: Upgrade to TensorFlow 2.11, 2.10.1, 2.9.3, or 2.8.4 — cherry-pick commit fc33f3dc4c14051a83eec6535b608abe1d355fde if running a custom build. 2. Short-term workaround: Add input validation middleware to inference endpoints that rejects requests with empty element_shape tensors before they reach the TF runtime. 3. Network hardening: Ensure TF Serving is not directly internet-exposed; place behind an API gateway or load balancer that performs basic schema validation. 4. Detection: Monitor for abnormal process restarts or segfault signals in TF Serving logs; alert on SIGSEGV in inference worker processes. 5. Verify: Run `python -c "import tensorflow as tf; print(tf.__version__)"` across all inference nodes to confirm patched versions.

What systems are affected by CVE-2022-41891?

This vulnerability affects the following AI/ML architecture patterns: model serving, training pipelines, inference, batch prediction pipelines.

What is the CVSS score for CVE-2022-41891?

CVE-2022-41891 has a CVSS v3.1 base score of 7.5 (HIGH). The EPSS exploitation probability is 0.43%.

What is the AI security impact?

Affected AI Architectures

model servingtraining pipelinesinferencebatch prediction pipelines

MITRE ATLAS Techniques

AML.T0010.001 AI Software

AML.T0029 Denial of AI Service

AML.T0049 Exploit Public-Facing Application

Compliance Controls Affected

EU AI Act: Art.15

ISO 42001: A.6.2.6

NIST AI RMF: GOVERN-1.7, MANAGE-2.4

OWASP LLM Top 10: LLM09:2025

What are the technical details?

Original Advisory

TensorFlow is an open source platform for machine learning. If `tf.raw_ops.TensorListConcat` is given `element_shape=[]`, it results segmentation fault which can be used to trigger a denial of service attack. We have patched the issue in GitHub commit fc33f3dc4c14051a83eec6535b608abe1d355fde. The fix will be included in TensorFlow 2.11. We will also cherrypick this commit on TensorFlow 2.10.1, 2.9.3, and TensorFlow 2.8.4, as these are also affected and still in supported range.

Exploitation Scenario

An adversary identifies a TensorFlow Serving endpoint via passive recon (e.g., exposed gRPC port 8500 or REST port 8501). They craft a PredictRequest invoking a model that uses TensorListConcat — or directly call `tf.raw_ops.TensorListConcat` via a SavedModel with a crafted `element_shape=[]` tensor. The single request triggers a segfault in the TF C++ kernel, crashing the serving process. If the service uses Kubernetes with liveness probes, the pod restarts and the attacker repeats — creating a persistent DoS loop. No credentials, no ML knowledge, no prior access required.

Weaknesses (CWE)

CWE-20 Improper Input Validation

CWE-20 — Improper Input Validation: The product receives input or data, but it does not validate or incorrectly validates that the input has the properties that are required to process the data safely and correctly.

[Architecture and Design] Consider using language-theoretic security (LangSec) techniques that characterize inputs using a formal language and build "recognizers" for that language. This effectively requires parsing to be a distinct layer that effectively enforces a boundary between raw input and internal data representations, instead of allowing parser code to be scattered throughout the program, where it could be subject to errors or inconsistencies that create weaknesses. [REF-1109] [REF-1110] [REF-1111]
[Architecture and Design] Use an input validation framework such as Struts or the OWASP ESAPI Validation API. Note that using a framework does not automatically address all input validation problems; be mindful of weaknesses that could arise from misusing the framework itself (CWE-1173).

Source: MITRE CWE corpus.