CriticalVulnerability

CVE-2025-49655: Keras deserialization flaw allows code execution via malicious files

Source: NVD(opens in a new tab)
Identifier: CVE-2025-49655
Published: October 17, 2025
Record updated: October 9, 2026

Affected: keras >= 3.11.0, < 3.11.3
Fixed in: 3.11.3
released August 21, 2025, 57 days before disclosure
Known exploitation: Not listed in the CISA Known Exploited Vulnerabilities catalog at the last check.
EPSS: 0.8%

Summary

CVE-2025-49655 affects the Keras framework in versions 3.11.0 up to but not including 3.11.3. A maliciously uploaded Keras file containing a TorchModuleWrapper class can run arbitrary code on an end user's system when loaded, even with safe mode enabled. The flaw (CWE-502) can be triggered through both local and remote files.

Mitigation

Fixed in 3.11.3 (upgrade Keras to version 3.11.3 or later). Reference: https://github.com/keras-team/keras/pull/21575

Details

Scores and exploitation

CVSS score: 9.8(critical)
Patch available: Yes
EPSS: 0.8%Estimated probability that this CVE is exploited in the next 30 days, from the FIRST Exploit Prediction Scoring System. Read on November 4, 2026.

Weaknesses and techniques

Weakness (CWE): CWE-502
Attack pattern (CAPEC): CAPEC-586
AI threat technique (MITRE ATLAS): AML.T0010

Affected

Packages: keras@>= 3.11.0, < 3.11.3 (fixed: 3.11.3)

Classification

Attack type: Supply chain
Attack sophistication: Moderate
AI component targeted: Framework
Impact: Confidentiality,
Integrity,
Availability
Labels: security

How this record was produced

Summaries, mitigations and labels on this page are generated automatically from the source. The linked source is authoritative. The source gives this advisory a description and no title, so the title above is a headline generated from the source text.

Original source: https://nvd.nist.gov/vuln/detail/CVE-2025-49655

First tracked: February 15, 2026 at 08:42 PM ET, from NVD/CVE Database

Classifier: claude-haiku-5-5, prompt v4

Classifier confidence: 95% (self-assessed)

Affected packages in the Exposure Registry

Matched by package name and ecosystem. Each entry shows whether the package delegates to a language model and how many tracked packages depend on it.

kerasPyPINo LLM dependency detected · 6 tracked dependents

Summary

Mitigation

Affected packages in the Exposure Registry

Related items