CVE-2023-25665 — HIGH (CVSS 7.5) AI Security Vulnerability

Q: Is CVE-2023-25665 actively exploited?

Proof-of-concept exploit code is publicly available for CVE-2023-25665, increasing the risk of exploitation.

Q: How to fix CVE-2023-25665?

1. PATCH: Upgrade to TensorFlow 2.12.0 or 2.11.1 — the only complete fix. 2. WORKAROUND: Add input validation before SparseSparseMaximum calls to verify tensor indices/values are non-null and within valid ranges. 3. NETWORK CONTROLS: Restrict inference API access to authenticated, authorized clients only; use API gateway rate-limiting to reduce DoS blast radius. 4. DETECTION: Monitor for repeated process crashes or pod restarts in ML serving infrastructure; alert on abnormal TensorFlow exception rates. 5. AUDIT: Identify all internal services consuming TF versions < 2.11.1 via dependency scanning (pip list, requirements.txt audit).

Q: What systems are affected by CVE-2023-25665?

This vulnerability affects the following AI/ML architecture patterns: model serving, training pipelines, ML API endpoints.

Q: What is the CVSS score for CVE-2023-25665?

CVE-2023-25665 has a CVSS v3.1 base score of 7.5 (HIGH). The EPSS exploitation probability is 0.13%.

CISO Take

Any TensorFlow deployment exposing model endpoints that accept sparse tensor inputs is vulnerable to unauthenticated remote crash — effectively a DoS with zero prerequisites. Patch to TF 2.12.0 or 2.11.1 immediately; if patching is delayed, restrict network access to inference endpoints and validate tensor inputs at the API gateway layer. Not actively exploited in the wild, but trivial to trigger.

Risk Assessment

High severity (CVSS 7.5) with network-accessible, zero-interaction, zero-privilege attack vector makes this readily exploitable against exposed TensorFlow serving endpoints. Impact is limited to availability — no data exfiltration or code execution path. Real-world risk depends heavily on exposure: internal-only training clusters face minimal risk, while public-facing ML inference APIs are directly in scope. Not in CISA KEV and no public PoC exploitation reported, reducing urgency marginally.

Affected Systems

Package	Ecosystem	Vulnerable Range	Patched
tensorflow	pip	—	No patch
195.0K OpenSSF 7.2 3.7K dependents Pushed today 4% patched ~1372d to patch Full package profile →

Do you use tensorflow? You're affected.

Severity & Risk

CVSS 3.1

7.5 / 10

EPSS

0.1%

chance of exploitation in 30 days

Higher than 31% 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, CISA SSVC, EPSS, trickest/cve, and Nuclei templates.

Attack Surface

AV Network

AC Low

PR None

UI None

S Unchanged

C None

I None

A High

Recommended Action

5 steps

PATCH

Upgrade to TensorFlow 2.12.0 or 2.11.1 — the only complete fix.
WORKAROUND

Add input validation before SparseSparseMaximum calls to verify tensor indices/values are non-null and within valid ranges.
NETWORK CONTROLS

Restrict inference API access to authenticated, authorized clients only; use API gateway rate-limiting to reduce DoS blast radius.
DETECTION

Monitor for repeated process crashes or pod restarts in ML serving infrastructure; alert on abnormal TensorFlow exception rates.
AUDIT

Identify all internal services consuming TF versions < 2.11.1 via dependency scanning (pip list, requirements.txt audit).

CISA SSVC Assessment

Decision Track*

Exploitation poc

Automatable Yes

Technical Impact partial

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

Classification

DoS Framework Inference AML.T0010.001 - AI Software AML.T0029 - Denial of AI Service AML.T0049 - Exploit Public-Facing Application

Compliance Impact

This CVE is relevant to:

EU AI Act

Article 9 - Risk management system

ISO 42001

8.4 - AI system operation and monitoring

NIST AI RMF

GOVERN-6.1 - Policies for third-party AI risks MANAGE-2.2 - Mechanisms to respond to and recover from AI risks

OWASP LLM Top 10

LLM05:2025 - Supply Chain Vulnerabilities

Frequently Asked Questions

What is CVE-2023-25665?

Any TensorFlow deployment exposing model endpoints that accept sparse tensor inputs is vulnerable to unauthenticated remote crash — effectively a DoS with zero prerequisites. Patch to TF 2.12.0 or 2.11.1 immediately; if patching is delayed, restrict network access to inference endpoints and validate tensor inputs at the API gateway layer. Not actively exploited in the wild, but trivial to trigger.

Is CVE-2023-25665 actively exploited?

Proof-of-concept exploit code is publicly available for CVE-2023-25665, increasing the risk of exploitation.

How to fix CVE-2023-25665?

1. PATCH: Upgrade to TensorFlow 2.12.0 or 2.11.1 — the only complete fix. 2. WORKAROUND: Add input validation before SparseSparseMaximum calls to verify tensor indices/values are non-null and within valid ranges. 3. NETWORK CONTROLS: Restrict inference API access to authenticated, authorized clients only; use API gateway rate-limiting to reduce DoS blast radius. 4. DETECTION: Monitor for repeated process crashes or pod restarts in ML serving infrastructure; alert on abnormal TensorFlow exception rates. 5. AUDIT: Identify all internal services consuming TF versions < 2.11.1 via dependency scanning (pip list, requirements.txt audit).

What systems are affected by CVE-2023-25665?

This vulnerability affects the following AI/ML architecture patterns: model serving, training pipelines, ML API endpoints.

What is the CVSS score for CVE-2023-25665?

CVE-2023-25665 has a CVSS v3.1 base score of 7.5 (HIGH). The EPSS exploitation probability is 0.13%.

Technical Details

NVD Description

TensorFlow is an open source platform for machine learning. Prior to versions 2.12.0 and 2.11.1, when `SparseSparseMaximum` is given invalid sparse tensors as inputs, it can give a null pointer error. A fix is included in TensorFlow version 2.12 and version 2.11.1.

Exploitation Scenario

An adversary identifies a public-facing TensorFlow Serving endpoint (e.g., via Shodan for port 8501/8500, or through API documentation leakage). They craft a POST request to the predict API containing an intentionally malformed SparseSparseMaximum input — a sparse tensor with null or zero-length index arrays. TensorFlow dereferences the null pointer inside the kernel, triggering a segfault that crashes the serving process. In Kubernetes deployments, the crash loop can be sustained by repeating the request faster than the restart policy recovers, achieving sustained service disruption against ML-dependent applications such as fraud detection, content moderation, or recommendation engines.