CVE-2025-54952

Critical

Published: 08 August 2025

Published

08 August 2025

Modified

15 April 2026

KEV Added

—

Patch

—

CVSS Score 9.8 CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H

EPSS Score 0.0036 58.3th percentile

Risk Priority 20 60% EPSS · 20% KEV · 20% CVSS

Summary

CVE-2025-54952 is a critical-severity Integer Overflow to Buffer Overflow (CWE-680) vulnerability in Facebook (inferred from references). Its CVSS base score is 9.8 (Critical).

Operationally, exploitation aligns with the MITRE ATT&CK technique Exploit Public-Facing Application (T1190); ranked in the top 41.7% of CVEs by exploit likelihood; it is not currently listed in the CISA KEV catalog.

The strongest mitigations our analysis identified are NIST 800-53 SI-2 (Flaw Remediation) and SI-10 (Information Input Validation).

Threat & Defense at a Glance

What attackers do: exploitation maps to Exploit Public-Facing Application (T1190) and 1 other technique. What defenders deploy: see the NIST 800-53 controls recommended below.

Threat & Defense Details

Mitigating Controls (NIST 800-53 r5)AI

SI-2 Flaw Remediation good match

prevent

Directly requires timely remediation of the integer overflow vulnerability in ExecuTorch by applying the vendor-provided patch prior to commit 8f062d3f661e20bb19b24b767b9a9a46e8359f2b.

SI-10 Information Input Validation partial match

prevent

Mandates validation of ExecuTorch model inputs during loading to detect and reject malformed data that could trigger the integer overflow leading to under-allocation.

SI-16 Memory Protection partial match

prevent

Implements memory protection mechanisms such as address space layout randomization or guard pages to mitigate exploitation of code execution from the under-allocated memory regions.

MITRE ATT&CK Enterprise TechniquesAI

T1190 Exploit Public-Facing Application Initial Access

Adversaries may attempt to exploit a weakness in an Internet-facing host or system to initially access a network.

attack.mitre.org →

T1203 Exploitation for Client Execution Execution

Adversaries may exploit software vulnerabilities in client applications to execute code.

attack.mitre.org →

Why these techniques?

Integer overflow in remote model loading directly enables unauthenticated remote code execution (T1190 for public-facing ExecuTorch instances; T1203 for client/edge model execution).

Confidence: HIGH · MITRE ATT&CK Enterprise v18.1

NVD Description

An integer overflow vulnerability in the loading of ExecuTorch models can cause smaller-than-expected memory regions to be allocated, potentially resulting in code execution or other undesirable effects. This issue affects ExecuTorch prior to commit 8f062d3f661e20bb19b24b767b9a9a46e8359f2b.

Deeper analysisAI

CVE-2025-54952 is an integer overflow vulnerability (CWE-680) in the loading of ExecuTorch models, which can cause smaller-than-expected memory regions to be allocated. This may lead to code execution or other undesirable effects. The vulnerability affects ExecuTorch versions prior to commit 8f062d3f661e20bb19b24b767b9a9a46e8359f2b. It has a CVSS v3.1 base score of 9.8 (AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H), indicating critical severity due to its high impact on confidentiality, integrity, and availability.

Attackers can exploit this vulnerability remotely over the network with low complexity, requiring no privileges, authentication, or user interaction. Successful exploitation allows arbitrary code execution on the target system hosting the vulnerable ExecuTorch instance, potentially compromising the entire environment where ExecuTorch models are loaded and executed.

Mitigation is available via the fixing commit at https://github.com/pytorch/executorch/commit/8f062d3f661e20bb19b24b767b9a9a46e8359f2b. Additional details are provided in the Meta/Facebook security advisory at https://www.facebook.com/security/advisories/cve-2025-54952. Security practitioners should update to the patched version and validate ExecuTorch model loading in AI/ML inference pipelines on edge or embedded devices.

This vulnerability is particularly relevant to AI/ML deployments, as ExecuTorch enables optimized execution of PyTorch models outside traditional server environments. No public reports of real-world exploitation were available as of the CVE publication on 2025-08-08.