9.8

CVE-2025-49655: keras Deserialization of Untrusted Data

CVSS: 9.8v3.1CriticalDeferred
AliasesGHSA-cvhh-q5g5-qprpPYSEC-2026-368UBUNTU-CVE-2025-49655
Published 17-10-2025Last modified 17-10-2025

Vulnerability summary

CVE-2025-49655 is a critical (CVSS 9.8) Deserialization of Untrusted Data vulnerability in keras. EPSS estimates a 0.75% chance of exploitation in the next 30 days, higher than 54% of scored vulnerabilities.

Overall risk
70/100
0100
High · default modelHow is this calculated?
CVSS score
9.8/10
010
Critical · v3.1 (CVE.org)
EPSS score
0.75%
0%100%
Low · 0.00 pp in 30 days
KEVNot listedIn no KEV catalog
RansomwareNone recordedNo known use recorded
Public exploits0No public exploits recorded

Timeline

  1. 17-10-2025
    PublishedVulnerability first disclosed.
  2. 17-10-2025
    Last modifiedVulnerability information updated.

Description

Deserialization of untrusted data can occur in versions of the Keras framework running versions 3.11.0 up to but not including 3.11.3, enabling a maliciously uploaded Keras file containing a TorchModuleWrapper class to run arbitrary code on an end user’s system when loaded despite safe mode being enabled. The vulnerability can be triggered through both local and remote files.

CVSS metrics

5 scores · 5 sources · 1 version

EPSS trend

Probability

0.75%

Chance of exploitation in the next 30 days

Percentile

54%

Versus all other CVEs

30-day change

0.00 pp

From 0.75% on 18-09-2026

18-09-202618-10-2026
EPSS score
0.75%

Known exploited

Compare KEV catalogs

Not in any of the 3 tracked catalogs.

Public exploits

0 found

No public exploits indexed.

Weaknesses

CWE · root cause

Affected systems

Debian1 product
  • kerasAll versions (No fix recorded)
Keras1 product
  • keras≥ 3.11.0, < 3.11.3
PyPI1 product
  • keras≥ 3.11.0, < 3.11.3
Ubuntu1 product
  • kerasAll versions (No fix recorded)

4 vendors · 4 products

References

10 links