CVE-2022-29196: TensorFlow DoS — MEDIUM

CISO Take

Low-priority patching item for teams running TensorFlow below 2.6.4/2.7.2/2.8.1/2.9.0. An attacker with local access can crash training jobs by passing a malformed filter_sizes argument to Conv3DBackpropFilterV2, triggering a CHECK assertion failure. Upgrade to any patched version; highest risk on multi-tenant shared GPU clusters where job isolation is weak.

What is the risk?

Low operational risk for most environments. Exploitation requires local access with user-level privileges — no remote vector exists without a prior foothold. Impact is strictly availability (no confidentiality or integrity exposure). Primary threat actor is a malicious insider or an adversary who has already compromised a training node or shared notebook environment. Not in CISA KEV; no public evidence of active exploitation.

What systems are affected?

Package	Ecosystem	Vulnerable Range	Patched
TensorFlow	pip	—	No patch
195.8K OpenSSF 7.1 3.7K dependents Pushed 2d ago 4% patched ~1372d to patch Full package profile →

Do you use TensorFlow? You're affected.

How severe is it?

CVSS 3.1

5.5 / 10

EPSS

0.3%

chance of exploitation in 30 days

Higher than 23% 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 Local

AC Low

PR Low

UI None

S Unchanged

C None

I None

A High

What should I do?

4 steps

Patch: Upgrade TensorFlow to ≥2.6.4, ≥2.7.2, ≥2.8.1, or ≥2.9.0 immediately if running 3D CNN workloads.
Isolation: Enforce job-level sandboxing on shared ML compute clusters; restrict who can submit arbitrary training scripts.
Detection: Monitor TensorFlow process logs for CHECK-failure stack traces in conv_grad_ops_3d.cc as an anomaly indicator.
Interim workaround if patching is delayed: validate that filter_sizes is a 1D tensor before calling the op in any custom training code.

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.T0029 - Denial of AI Service AML.T0034 - Cost Harvesting 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

ISO 42001

A.9.7 - AI system robustness and reliability

NIST AI RMF

MANAGE 2.2 - Resilience and reliability of AI systems

OWASP LLM Top 10

LLM04 - Model Denial of Service

Frequently Asked Questions

What is CVE-2022-29196?

Low-priority patching item for teams running TensorFlow below 2.6.4/2.7.2/2.8.1/2.9.0. An attacker with local access can crash training jobs by passing a malformed filter_sizes argument to Conv3DBackpropFilterV2, triggering a CHECK assertion failure. Upgrade to any patched version; highest risk on multi-tenant shared GPU clusters where job isolation is weak.

Is CVE-2022-29196 actively exploited?

Proof-of-concept exploit code is publicly available for CVE-2022-29196, increasing the risk of exploitation.

How to fix CVE-2022-29196?

1. Patch: Upgrade TensorFlow to ≥2.6.4, ≥2.7.2, ≥2.8.1, or ≥2.9.0 immediately if running 3D CNN workloads. 2. Isolation: Enforce job-level sandboxing on shared ML compute clusters; restrict who can submit arbitrary training scripts. 3. Detection: Monitor TensorFlow process logs for CHECK-failure stack traces in conv_grad_ops_3d.cc as an anomaly indicator. 4. Interim workaround if patching is delayed: validate that filter_sizes is a 1D tensor before calling the op in any custom training code.

What systems are affected by CVE-2022-29196?

This vulnerability affects the following AI/ML architecture patterns: training pipelines, MLOps infrastructure, shared GPU compute clusters.

What is the CVSS score for CVE-2022-29196?

CVE-2022-29196 has a CVSS v3.1 base score of 5.5 (MEDIUM). The EPSS exploitation probability is 0.32%.

What is the AI security impact?

Affected AI Architectures

training pipelinesMLOps infrastructureshared GPU compute clusters

MITRE ATLAS Techniques

AML.T0029 Denial of AI Service

AML.T0034 Cost Harvesting

AML.T0049 Exploit Public-Facing Application

Compliance Controls Affected

EU AI Act: Art. 15

ISO 42001: A.9.7

NIST AI RMF: MANAGE 2.2

OWASP LLM Top 10: LLM04

What are the technical details?

Original Advisory

TensorFlow is an open source platform for machine learning. Prior to versions 2.9.0, 2.8.1, 2.7.2, and 2.6.4, the implementation of `tf.raw_ops.Conv3DBackpropFilterV2` does not fully validate the input arguments. This results in a `CHECK`-failure which can be used to trigger a denial of service attack. The code does not validate that the `filter_sizes` argument is a vector. Versions 2.9.0, 2.8.1, 2.7.2, and 2.6.4 contain a patch for this issue.

Exploitation Scenario

An attacker with access to a shared ML training platform — internal GPU cluster, cloud notebook (Vertex AI, SageMaker), or CI/CD pipeline running model training — submits a script calling tf.raw_ops.Conv3DBackpropFilterV2 with filter_sizes passed as a 2D tensor instead of a vector. TensorFlow's CHECK macro fires, immediately killing the training process with a SIGABRT. On a shared cluster, this crashes co-located jobs and can be looped to continuously deny training capacity, delaying model delivery or erasing unsaved checkpoints from production model retraining pipelines.

Weaknesses (CWE)

CWE-1284 Improper Validation of Specified Quantity in Input Primary CWE-20 Improper Input Validation

CWE-1284 — Improper Validation of Specified Quantity in Input: The product receives input that is expected to specify a quantity (such as size or length), but it does not validate or incorrectly validates that the quantity has the required properties.

[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

Source: MITRE CWE corpus.