CVE-2022-35983: TensorFlow DoS — HIGH

CISO Take

A remotely exploitable denial-of-service in TensorFlow's Save and SaveSlices ops allows any unauthenticated attacker to crash TensorFlow processes by supplying tensors with unsupported dtypes. No workaround exists — patch to TF 2.10.0, 2.9.1, 2.8.1, or 2.7.2 immediately. Any ML serving infrastructure accepting user-controlled tensor inputs is directly exposed.

What is the risk?

CVSS 7.5 (AV:N/AC:L/PR:N/UI:N) reflects a reliably exploitable, no-auth-required crash. The reachable assertion (CWE-617) is deterministic: a single malformed request terminates the process. Exposure is highest in model-serving and training orchestration layers where TensorFlow operations are reachable via network — common in production MLOps stacks. Not in CISA KEV and no active exploitation evidence, but trivial to trigger once the target is identified.

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 30% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

Exploit Available

Exploitation: MEDIUM

Sophistication

Trivial

Exploitation Confidence

medium

○ 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 (or cherrypick commits to 2.9.1, 2.8.1, 2.7.2). No workaround exists per the advisory.
Input validation: enforce dtype allowlists at the application boundary before tensors reach Save/SaveSlices — reject or coerce unsupported dtypes.
Process resilience: configure auto-restart policies (Kubernetes restartPolicy: Always, systemd Restart=on-failure) to minimize downtime if exploited.
Detection: alert on repeated TensorFlow process crashes (OOMKilled or non-zero exit codes) in serving pods — abnormal crash rates may indicate probing.
Network segmentation: ensure TF Serving gRPC/REST endpoints are not public-facing without an authenticated API gateway.

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.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 (high-risk AI systems)

ISO 42001

A.9.1 - Availability of AI system

NIST AI RMF

MANAGE-2.2 - Mechanisms to sustain the value of AI systems are established

OWASP LLM Top 10

LLM05:2025 - Improper Output Handling / Supply Chain Vulnerabilities

Frequently Asked Questions

What is CVE-2022-35983?

A remotely exploitable denial-of-service in TensorFlow's Save and SaveSlices ops allows any unauthenticated attacker to crash TensorFlow processes by supplying tensors with unsupported dtypes. No workaround exists — patch to TF 2.10.0, 2.9.1, 2.8.1, or 2.7.2 immediately. Any ML serving infrastructure accepting user-controlled tensor inputs is directly exposed.

Is CVE-2022-35983 actively exploited?

Proof-of-concept exploit code is publicly available for CVE-2022-35983, increasing the risk of exploitation.

How to fix CVE-2022-35983?

1. Patch: upgrade to TensorFlow 2.10.0 (or cherrypick commits to 2.9.1, 2.8.1, 2.7.2). No workaround exists per the advisory. 2. Input validation: enforce dtype allowlists at the application boundary before tensors reach Save/SaveSlices — reject or coerce unsupported dtypes. 3. Process resilience: configure auto-restart policies (Kubernetes restartPolicy: Always, systemd Restart=on-failure) to minimize downtime if exploited. 4. Detection: alert on repeated TensorFlow process crashes (OOMKilled or non-zero exit codes) in serving pods — abnormal crash rates may indicate probing. 5. Network segmentation: ensure TF Serving gRPC/REST endpoints are not public-facing without an authenticated API gateway.

What systems are affected by CVE-2022-35983?

This vulnerability affects the following AI/ML architecture patterns: model serving, training pipelines, ML infrastructure, federated learning nodes.

What is the CVSS score for CVE-2022-35983?

CVE-2022-35983 has a CVSS v3.1 base score of 7.5 (HIGH). The EPSS exploitation probability is 0.38%.

What is the AI security impact?

Affected AI Architectures

model servingtraining pipelinesML infrastructurefederated learning nodes

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.9.1

NIST AI RMF: MANAGE-2.2

OWASP LLM Top 10: LLM05:2025

What are the technical details?

Original Advisory

TensorFlow is an open source platform for machine learning. If `Save` or `SaveSlices` is run over tensors of an unsupported `dtype`, it results in a `CHECK` fail that can be used to trigger a denial of service attack. We have patched the issue in GitHub commit 5dd7b86b84a864b834c6fa3d7f9f51c87efa99d4. 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 TensorFlow Serving deployment (e.g., via Shodan scan for gRPC port 8500 or REST port 8501). They submit a PredictRequest with a tensor of an unsupported dtype (e.g., DT_RESOURCE or DT_VARIANT) targeting a model that internally invokes Save or SaveSlices. The operation hits a CHECK assertion, the TF Serving process terminates with a fatal log, and inference for all models on that instance is interrupted. With no auto-restart, SLA is breached. An attacker can loop this request to defeat restart policies and maintain a persistent DoS against the ML inference layer.

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.