CVE-2022-36027: TensorFlow DoS crash

CISO Take

A remote, unauthenticated attacker can crash any Python process running TensorFlow's model converter by triggering the vulnerable per-channel quantization path for transposed convolutions — no privileges required. If your MLOps pipeline exposes model conversion as a service or accepts external model configs, this is a denial-of-service risk with CVSS 7.5. Patch immediately to TF 2.10.0 / 2.9.1 / 2.8.1 / 2.7.2; there are no workarounds.

What is the risk?

Risk is elevated for organizations running TensorFlow conversion pipelines that accept external or untrusted model inputs. The CVSS 7.5 reflects low attack complexity, no authentication required, and full availability impact (process crash). However, impact is limited to availability — no data exfiltration or code execution. Real-world risk is highest in automated model optimization services, CI/CD pipelines that convert third-party models, and inference platforms that perform on-the-fly quantization. Not in CISA KEV and no known active exploitation, which tempers urgency slightly.

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

0.6%

chance of exploitation in 30 days

Higher than 43% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

Exploit Available

Exploitation: MEDIUM

Sophistication

Moderate

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 2.10.0, 2.9.1 (TF 2.9.x), 2.8.1 (TF 2.8.x), or 2.7.2 (TF 2.7.x). Verify via pip show tensorflow | grep Version.
Workaround

None available per TensorFlow advisory — patching is mandatory.
Input validation

If accepting external model files for conversion, add a model schema validation layer before invoking the converter; reject models with transposed convolution layers if patching is delayed.
Process isolation

Run model conversion jobs in isolated containers or subprocesses with resource limits and automatic restart — limits blast radius of crash to the conversion job, not the host service.
Detection

Monitor for unexpected TensorFlow process exits (exit code non-zero) in conversion pipelines; alert on repeated crashes from the same source IP/model submission.

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 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

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

ISO 42001

A.10.1 - Information security in supplier relationships

NIST AI RMF

MANAGE 2.4 - Residual risks to individuals, groups, communities, organizations, and society are managed and documented

Frequently Asked Questions

What is CVE-2022-36027?

A remote, unauthenticated attacker can crash any Python process running TensorFlow's model converter by triggering the vulnerable per-channel quantization path for transposed convolutions — no privileges required. If your MLOps pipeline exposes model conversion as a service or accepts external model configs, this is a denial-of-service risk with CVSS 7.5. Patch immediately to TF 2.10.0 / 2.9.1 / 2.8.1 / 2.7.2; there are no workarounds.

Is CVE-2022-36027 actively exploited?

Proof-of-concept exploit code is publicly available for CVE-2022-36027, increasing the risk of exploitation.

How to fix CVE-2022-36027?

1. **Patch**: Upgrade to TensorFlow 2.10.0, 2.9.1 (TF 2.9.x), 2.8.1 (TF 2.8.x), or 2.7.2 (TF 2.7.x). Verify via `pip show tensorflow | grep Version`. 2. **Workaround**: None available per TensorFlow advisory — patching is mandatory. 3. **Input validation**: If accepting external model files for conversion, add a model schema validation layer before invoking the converter; reject models with transposed convolution layers if patching is delayed. 4. **Process isolation**: Run model conversion jobs in isolated containers or subprocesses with resource limits and automatic restart — limits blast radius of crash to the conversion job, not the host service. 5. **Detection**: Monitor for unexpected TensorFlow process exits (exit code non-zero) in conversion pipelines; alert on repeated crashes from the same source IP/model submission.

What systems are affected by CVE-2022-36027?

This vulnerability affects the following AI/ML architecture patterns: training pipelines, model serving, MLOps pipelines, edge/mobile deployment pipelines.

What is the CVSS score for CVE-2022-36027?

CVE-2022-36027 has a CVSS v3.1 base score of 7.5 (HIGH). The EPSS exploitation probability is 0.58%.

What is the AI security impact?

Affected AI Architectures

training pipelinesmodel servingMLOps pipelinesedge/mobile deployment 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: Article 15, Article 9

ISO 42001: A.10.1

NIST AI RMF: MANAGE 2.4

What are the technical details?

Original Advisory

TensorFlow is an open source platform for machine learning. When converting transposed convolutions using per-channel weight quantization the converter segfaults and crashes the Python process. We have patched the issue in GitHub commit aa0b852a4588cea4d36b74feb05d93055540b450. The fix will be included in TensorFlow 2.10.0. We will also cherrypick this commit on TensorFlow 2.9.1, TensorFlow 2.8.1, and TensorFlow 2.7.2, as these are also affected and still in supported range. There are no known workarounds for this issue.

Exploitation Scenario

An adversary targeting an MLOps platform that exposes a model optimization API (e.g., 'submit a TF SavedModel, receive a quantized TFLite file') crafts a model containing transposed convolution layers with per-channel weight quantization configured. Submitting this model to the conversion endpoint causes the TensorFlow converter to segfault, killing the backend Python worker. By submitting the malformed model repeatedly, the adversary can maintain a persistent denial-of-service against the conversion service, blocking legitimate model deployment and disrupting AI product pipelines. No authentication, no payload — just a structurally valid but pathological model file.

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.