CVE-2022-41893: TensorFlow: DoS via TensorListResize malformed input

HIGHPoC AVAILABLECISA: TRACK*

Published November 18, 2022

CISO Take

A remotely exploitable denial-of-service in TensorFlow's TensorListResize op allows any unauthenticated attacker to crash your model serving infrastructure by sending a nonscalar 'size' input. No authentication, no user interaction — a single malformed request is sufficient. Patch immediately to TF 2.11, 2.10.1, 2.9.3, or 2.8.4; if you cannot patch, add input shape validation at your API gateway layer.

What is the risk?

High risk for organizations exposing TensorFlow-based inference endpoints without input validation. CVSS 7.5 with network vector, low complexity, no privileges, no user interaction — the attack profile is trivial. The CHECK fail terminates the TF process, making this a reliable crasher for automated attacks. Not in CISA KEV and no public weaponization reported, but the simplicity of exploitation elevates practical risk. Organizations running TF 2.8–2.10 in production model serving without perimeter controls are directly exposed.

What systems are affected?

Package	Ecosystem	Vulnerable Range	Patched
TensorFlow	pip	—	No patch
201.3KOpenSSF 7.63.4K dependentsPushed 6d ago4% patched~1372d to patchFull package profile →

Do you use TensorFlow? You're affected.

How severe is it?

CVSS 3.1

7.5 / 10

EPSS

0.5%

chance of exploitation in 30 days

Higher than 43% 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?

AVNetwork

ACLow

PRNone

UINone

SUnchanged

CNone

INone

AHigh

What should I do?

1 step

1) PATCH: Upgrade to TF 2.11, 2.10.1, 2.9.3, or 2.8.4 (cherry-picked fix in each). Commit 888e34b49009a4e734c27ab0c43b0b5102682c56 is the authoritative fix. 2) WORKAROUND: Validate that all 'size' inputs to TensorListResize are scalar (rank-0) tensors before execution; reject nonscalar inputs at the application layer. 3) PERIMETER: Place API gateways or input validation middleware in front of TF Serving endpoints to reject malformed tensor shapes. 4) DETECT: Alert on unexpected TF process restarts or CHECK fail messages in serving logs (grep 'Check failed' in TF stderr). 5) ISOLATE: Run TF Serving in containers with restart policies; this won't prevent the DoS but limits blast radius and speeds recovery.

What does CISA's SSVC say?

DecisionTrack*

Exploitationpoc

AutomatableNo

Technical Impactpartial

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

How is it classified?

DoSFrameworkInferenceAML.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

Article 15-Accuracy, robustness and cybersecurity

ISO 42001

A.8.4-AI system operation and monitoring

NIST AI RMF

MANAGE 2.2-Mechanisms to sustain and monitor AI risk treatments

OWASP LLM Top 10

LLM04-Model Denial of Service

Frequently Asked Questions

What is CVE-2022-41893?

Is CVE-2022-41893 actively exploited?

Proof-of-concept exploit code is publicly available for CVE-2022-41893, increasing the risk of exploitation.

How to fix CVE-2022-41893?

1) PATCH: Upgrade to TF 2.11, 2.10.1, 2.9.3, or 2.8.4 (cherry-picked fix in each). Commit 888e34b49009a4e734c27ab0c43b0b5102682c56 is the authoritative fix. 2) WORKAROUND: Validate that all 'size' inputs to TensorListResize are scalar (rank-0) tensors before execution; reject nonscalar inputs at the application layer. 3) PERIMETER: Place API gateways or input validation middleware in front of TF Serving endpoints to reject malformed tensor shapes. 4) DETECT: Alert on unexpected TF process restarts or CHECK fail messages in serving logs (grep 'Check failed' in TF stderr). 5) ISOLATE: Run TF Serving in containers with restart policies; this won't prevent the DoS but limits blast radius and speeds recovery.

What systems are affected by CVE-2022-41893?

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

What is the CVSS score for CVE-2022-41893?

CVE-2022-41893 has a CVSS v3.1 base score of 7.5 (HIGH). The EPSS exploitation probability is 0.49%.

What is the AI security impact?

Affected AI Architectures

model servinginference APIstraining pipelinesbatch inference jobs

MITRE ATLAS Techniques

AML.T0029Denial of AI Service

AML.T0034Cost Harvesting

AML.T0049Exploit Public-Facing Application

Compliance Controls Affected

EU AI Act: Article 15

ISO 42001: A.8.4

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. If `tf.raw_ops.TensorListResize` is given a nonscalar value for input `size`, it results `CHECK` fail which can be used to trigger a denial of service attack. We have patched the issue in GitHub commit 888e34b49009a4e734c27ab0c43b0b5102682c56. 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 an organization's publicly accessible TF Serving gRPC or REST endpoint. They craft a PredictRequest with a TensorListResize call where the 'size' argument is a 2D tensor instead of a scalar. TensorFlow performs no shape validation before the operation, triggering an internal CHECK assertion failure that immediately terminates the serving process. The attacker automates this request in a loop — requiring no credentials, no session, and no ML expertise — keeping the inference endpoint permanently unavailable. In Kubernetes deployments, the constant crash-restart cycle exhausts pod restart budgets and causes broader service disruption.

Weaknesses (CWE)

CWE-617Reachable Assertion

CWE-617 — Reachable Assertion: The product contains an assert() or similar statement that can be triggered by an attacker, which leads to an application exit or other behavior that is more severe than necessary.

[Implementation] If the assertion statement is required (e.g., for security or reliability), change the invoking code to avoid the conditions that trigger the assertion.
[Implementation] Consider changing the assertion to some other less "fatal" error. This might not be easy to implement, since other code may be needed to handle the error.

Source: MITRE CWE corpus.