CVE-2022-35984: TensorFlow int64 type mismatch

CISO Take

A remotely exploitable denial-of-service in TensorFlow's ParameterizedTruncatedNormal op allows any unauthenticated attacker to crash a model serving endpoint by passing an int64 shape argument where int32 is expected. No authentication or ML knowledge is required — this is trivially automatable. Patch immediately to TF 2.10.0, 2.9.1, 2.8.1, or 2.7.2; any externally-facing TensorFlow inference service on older versions is exposed.

What is the risk?

High availability risk for any network-exposed TensorFlow serving infrastructure. CVSS 7.5 reflects realistic exploit ease: zero privileges, no user interaction, low complexity over the network. The blast radius is limited to DoS — no code execution or data exfiltration. However, for production AI systems (fraud detection, content moderation, recommendation engines), service disruption has direct business impact. Risk is highest for teams running unpatched TF versions behind API gateways with insufficient input type validation.

What systems are affected?

Package	Ecosystem	Vulnerable Range	Patched
TensorFlow	pip	—	No patch
195.8K OpenSSF 7.1 3.7K dependents Pushed 4d 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 TF 2.10.0, 2.9.1, 2.8.1, or 2.7.2 — the fix is in commit 72180be.
VALIDATE INPUTS

Enforce strict dtype checking on shape inputs at the API boundary before they reach TF ops; reject int64 where int32 is required.
ISOLATE

Run TF serving processes in isolated containers/VMs so a crash does not cascade; implement auto-restart with circuit breakers.
MONITOR

Alert on abnormal process crashes or CHECK failure signatures in TF serving logs.
AUDIT EXPOSURE

Inventory all TF versions in production and CI/CD pipelines — legacy training environments are often overlooked.

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.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 for high-risk AI

ISO 42001

A.9.3 - AI system robustness and resilience

NIST AI RMF

MANAGE 2.2 - Mechanisms to sustain the value of AI systems

OWASP LLM Top 10

LLM10:2025 - Unbounded Consumption

Frequently Asked Questions

What is CVE-2022-35984?

A remotely exploitable denial-of-service in TensorFlow's ParameterizedTruncatedNormal op allows any unauthenticated attacker to crash a model serving endpoint by passing an int64 shape argument where int32 is expected. No authentication or ML knowledge is required — this is trivially automatable. Patch immediately to TF 2.10.0, 2.9.1, 2.8.1, or 2.7.2; any externally-facing TensorFlow inference service on older versions is exposed.

Is CVE-2022-35984 actively exploited?

Proof-of-concept exploit code is publicly available for CVE-2022-35984, increasing the risk of exploitation.

How to fix CVE-2022-35984?

1. PATCH: Upgrade to TF 2.10.0, 2.9.1, 2.8.1, or 2.7.2 — the fix is in commit 72180be. 2. VALIDATE INPUTS: Enforce strict dtype checking on shape inputs at the API boundary before they reach TF ops; reject int64 where int32 is required. 3. ISOLATE: Run TF serving processes in isolated containers/VMs so a crash does not cascade; implement auto-restart with circuit breakers. 4. MONITOR: Alert on abnormal process crashes or CHECK failure signatures in TF serving logs. 5. AUDIT EXPOSURE: Inventory all TF versions in production and CI/CD pipelines — legacy training environments are often overlooked.

What systems are affected by CVE-2022-35984?

This vulnerability affects the following AI/ML architecture patterns: model serving, training pipelines, ML platform multi-tenant environments, probabilistic/generative model endpoints.

What is the CVSS score for CVE-2022-35984?

CVE-2022-35984 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 platform multi-tenant environmentsprobabilistic/generative model endpoints

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

ISO 42001: A.9.3

NIST AI RMF: MANAGE 2.2

OWASP LLM Top 10: LLM10:2025

What are the technical details?

Original Advisory

TensorFlow is an open source platform for machine learning. `ParameterizedTruncatedNormal` assumes `shape` is of type `int32`. A valid `shape` of type `int64` results in a mismatched type `CHECK` fail that can be used to trigger a denial of service attack. We have patched the issue in GitHub commit 72180be03447a10810edca700cbc9af690dfeb51. 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 targets a public-facing ML inference API (e.g., a recommendation or fraud-scoring service built on TF Serving). They send a malformed prediction request with a shape tensor of dtype int64 instead of the expected int32 to an endpoint that internally calls ParameterizedTruncatedNormal (common in probabilistic models and generative components). The mismatched type triggers a C++ CHECK assertion failure, crashing the TF serving worker process. With no rate limiting, the attacker automates this in a loop to sustain a denial-of-service condition, effectively taking down the ML service. No credentials, no prior access, and no ML expertise required — a basic fuzzing tool discovers the payload.

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.