CVE-2025-55551: PyTorch DoS in linalg.lu

CISO Take

Any PyTorch-based inference service exposed to untrusted input is at risk of remote crash with no authentication required. If your ML serving infrastructure uses torch.linalg.lu (common in scientific and financial ML pipelines), treat this as an urgent patching item. Until v2.8.0 is patched, firewall PyTorch endpoints from public access or add input shape validation at the API layer.

What is the risk?

HIGH operational risk for organizations running PyTorch in production serving roles. CVSS 7.5 is backed by AV:N/AC:L/PR:N/UI:N — an unauthenticated remote attacker with network reach to a PyTorch-backed service can crash it with a single crafted request. No exploit complexity, no foothold required. PoC exists publicly (GitHub gist referenced in NVD). Not in CISA KEV yet, but the low barrier to exploitation means weaponization is trivial.

What systems are affected?

Package	Ecosystem	Vulnerable Range	Patched
PyTorch	pip	—	No patch
100.9K OpenSSF 6.4 22.7K dependents Pushed 3d ago 11% patched ~216d to patch Full package profile →

Do you use PyTorch? You're affected.

How severe is it?

CVSS 3.1

7.5 / 10

EPSS

0.4%

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

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

Pin PyTorch to a version prior to 2.8.0 or await official patch — no patched version is confirmed yet; monitor https://github.com/pytorch/pytorch/issues/151401 for resolution.
WORKAROUND

Add input validation before any torch.linalg.lu call — validate tensor shape, dtype, and dimension constraints at the API boundary.
NETWORK

Restrict PyTorch inference endpoints to authenticated, internal traffic only; block public exposure of linalg-capable endpoints.
DETECT

Monitor for abnormal process termination or OOM events on model serving nodes; alert on repeated 5xx errors from inference endpoints.
ISOLATE

Run inference workloads in containerized environments with resource limits (memory/CPU cgroups) to contain blast radius of DoS.

What does CISA's SSVC say?

Decision Track*

Exploitation poc

Automatable Yes

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.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

6.1.2 - AI risk assessment 8.4 - AI system operation and monitoring

NIST AI RMF

MANAGE-2.2 - Mechanisms to sustain AI system availability MAP-5.1 - Likelihood and magnitude of impacts from third-party dependencies

OWASP LLM Top 10

LLM04:2025 - Model Denial of Service

Frequently Asked Questions

What is CVE-2025-55551?

Any PyTorch-based inference service exposed to untrusted input is at risk of remote crash with no authentication required. If your ML serving infrastructure uses torch.linalg.lu (common in scientific and financial ML pipelines), treat this as an urgent patching item. Until v2.8.0 is patched, firewall PyTorch endpoints from public access or add input shape validation at the API layer.

Is CVE-2025-55551 actively exploited?

No confirmed active exploitation of CVE-2025-55551 has been reported, but organizations should still patch proactively.

How to fix CVE-2025-55551?

1. PATCH: Pin PyTorch to a version prior to 2.8.0 or await official patch — no patched version is confirmed yet; monitor https://github.com/pytorch/pytorch/issues/151401 for resolution. 2. WORKAROUND: Add input validation before any torch.linalg.lu call — validate tensor shape, dtype, and dimension constraints at the API boundary. 3. NETWORK: Restrict PyTorch inference endpoints to authenticated, internal traffic only; block public exposure of linalg-capable endpoints. 4. DETECT: Monitor for abnormal process termination or OOM events on model serving nodes; alert on repeated 5xx errors from inference endpoints. 5. ISOLATE: Run inference workloads in containerized environments with resource limits (memory/CPU cgroups) to contain blast radius of DoS.

What systems are affected by CVE-2025-55551?

This vulnerability affects the following AI/ML architecture patterns: model serving, inference APIs, training pipelines, ML platform (multi-tenant), scientific/numerical ML workloads.

What is the CVSS score for CVE-2025-55551?

CVE-2025-55551 has a CVSS v3.1 base score of 7.5 (HIGH). The EPSS exploitation probability is 0.39%.

What is the AI security impact?

Affected AI Architectures

model servinginference APIstraining pipelinesML platform (multi-tenant)scientific/numerical ML workloads

MITRE ATLAS Techniques

AML.T0010.001 AI Software

AML.T0029 Denial of AI Service

AML.T0034 Cost Harvesting

AML.T0049 Exploit Public-Facing Application

Compliance Controls Affected

EU AI Act: Article 15

ISO 42001: 6.1.2, 8.4

NIST AI RMF: MANAGE-2.2, MAP-5.1

OWASP LLM Top 10: LLM04:2025

What are the technical details?

Original Advisory

An issue in the component torch.linalg.lu of pytorch v2.8.0 allows attackers to cause a Denial of Service (DoS) when performing a slice operation.

Exploitation Scenario

Adversary identifies an exposed ML inference API (e.g., TorchServe or a FastAPI endpoint) that accepts matrix inputs for a scientific or financial ML model. They review the public PoC gist, craft a minimal HTTP POST with a tensor designed to trigger the malformed slice operation in torch.linalg.lu, and send it. The PyTorch worker crashes or hangs, taking down the inference service. Repeated requests prevent recovery. In a multi-tenant ML platform, this could deny service to all tenants sharing the PyTorch backend.

Weaknesses (CWE)

CWE-400 Uncontrolled Resource Consumption

CWE-400 — Uncontrolled Resource Consumption: The product does not properly control the allocation and maintenance of a limited resource.

[Architecture and Design] Design throttling mechanisms into the system architecture. The best protection is to limit the amount of resources that an unauthorized user can cause to be expended. A strong authentication and access control model will help prevent such attacks from occurring in the first place. The login application should be protected against DoS attacks as much as possible. Limiting the database access, perhaps by caching result sets, can help minimize the resources expended. To further limit the potential for a DoS attack, consider tracking the rate of requests received from users and blocking requests that exceed a defined rate threshold.
[Architecture and Design] Mitigation of resource exhaustion attacks requires that the target system either: The first of these solutions is an issue in itself though, since it may allow attackers to prevent the use of the system by a particular valid user. If the attacker impersonates the valid user, they may be able to prevent the user from accessing the server in question. The second solution is simply difficult to effectively institute -- and even when properly done, it does not provide a full solution. It simply makes the attack require more resources on the part of the attacker. recognizes the attack and denies that user further access for a given amount of time, or uniformly throttles all requests in order to make it more difficult to consume resources more quickly than they can again be freed.

Source: MITRE CWE corpus.