CVE-2022-36016: TensorFlow CHECK-fail assertion

CISO Take

A network-reachable assertion failure in TensorFlow's type inference allows unauthenticated attackers to crash any exposed TensorFlow serving process with a single malformed request. If you run TF Serving or any TensorFlow-based inference endpoint (including cloud-hosted), patch to TF 2.10.0 / 2.9.1 / 2.8.1 / 2.7.2 immediately. No workaround exists — patching is the only remediation.

What is the risk?

HIGH severity in practice for organizations exposing TensorFlow inference APIs. CVSS 7.5 with AV:N/AC:L/PR:N/UI:N means any internet-facing TF endpoint is reachable with no authentication and trivial effort. The availability-only impact (A:H) limits blast radius — no data exfiltration or code execution — but a persistent crash loop against a production ML serving cluster causes full service outage. Risk is amplified in AI-native organizations where TF Serving is a critical dependency in real-time inference pipelines.

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

0.5%

chance of exploitation in 30 days

Higher than 40% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

No known exploitation

Sophistication

Trivial

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, or apply cherrypick patches to 2.9.1, 2.8.1, or 2.7.2. Commit 6104f0d4091c260ce9352f9155f7e9b725eab012 is the fix.
ISOLATE

Place TF Serving instances behind API gateways with authentication — eliminate unauthenticated public access to inference endpoints.
RATE LIMIT

Deploy request rate limiting and input validation at the API gateway layer to reduce crash-loop risk while patching is in progress.
DETECT

Monitor TF Serving process crash rates and unexpected restarts; alert on repeated process exits within short windows.
INVENTORY

Audit all internal and external TF model serving endpoints to identify exposure surface before patching.

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?

DoS Framework Inference AML.T0029 - Denial of AI Service AML.T0040 - AI Model Inference API Access 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

ISO 42001

A.6.2.5 - AI system availability and resilience

NIST AI RMF

MANAGE 2.4 - Residual risks from AI systems are managed

OWASP LLM Top 10

LLM04 - Model Denial of Service

Frequently Asked Questions

What is CVE-2022-36016?

A network-reachable assertion failure in TensorFlow's type inference allows unauthenticated attackers to crash any exposed TensorFlow serving process with a single malformed request. If you run TF Serving or any TensorFlow-based inference endpoint (including cloud-hosted), patch to TF 2.10.0 / 2.9.1 / 2.8.1 / 2.7.2 immediately. No workaround exists — patching is the only remediation.

Is CVE-2022-36016 actively exploited?

No confirmed active exploitation of CVE-2022-36016 has been reported, but organizations should still patch proactively.

How to fix CVE-2022-36016?

1. PATCH: Upgrade to TensorFlow 2.10.0, or apply cherrypick patches to 2.9.1, 2.8.1, or 2.7.2. Commit 6104f0d4091c260ce9352f9155f7e9b725eab012 is the fix. 2. ISOLATE: Place TF Serving instances behind API gateways with authentication — eliminate unauthenticated public access to inference endpoints. 3. RATE LIMIT: Deploy request rate limiting and input validation at the API gateway layer to reduce crash-loop risk while patching is in progress. 4. DETECT: Monitor TF Serving process crash rates and unexpected restarts; alert on repeated process exits within short windows. 5. INVENTORY: Audit all internal and external TF model serving endpoints to identify exposure surface before patching.

What systems are affected by CVE-2022-36016?

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

What is the CVSS score for CVE-2022-36016?

CVE-2022-36016 has a CVSS v3.1 base score of 7.5 (HIGH). The EPSS exploitation probability is 0.52%.

What is the AI security impact?

Affected AI Architectures

model servingtraining pipelinesinference infrastructureMLOps pipelines

MITRE ATLAS Techniques

AML.T0029 Denial of AI Service

AML.T0040 AI Model Inference API Access

AML.T0049 Exploit Public-Facing Application

Compliance Controls Affected

EU AI Act: Art. 15

ISO 42001: A.6.2.5

NIST AI RMF: MANAGE 2.4

OWASP LLM Top 10: LLM04

What are the technical details?

Original Advisory

TensorFlow is an open source platform for machine learning. When `tensorflow::full_type::SubstituteFromAttrs` receives a `FullTypeDef& t` that is not exactly three args, it triggers a `CHECK`-fail instead of returning a status. We have patched the issue in GitHub commit 6104f0d4091c260ce9352f9155f7e9b725eab012. 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 identifies a production ML inference API powered by TensorFlow Serving (e.g., via HTTP headers, error messages, or open-source intelligence). Without authentication, they craft a malformed inference request that triggers the `SubstituteFromAttrs` code path with an unexpected number of type arguments. The TF process hits the CHECK-fail assertion and crashes. Automated process restarts (common in Kubernetes deployments) bring the service back up, allowing the attacker to repeat the crash in a loop — effectively maintaining a persistent denial of service against the ML inference layer. For organizations using AI models in fraud detection, content moderation, or real-time decisioning, this translates directly into business disruption.

Weaknesses (CWE)

CWE-617 Reachable 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] Make sensitive open/close operation non reachable by directly user-controlled data (e.g. open/close resources)
[Implementation] Perform input validation on user data.

Source: MITRE CWE corpus.