8.6

CVE-2025-8747: google keras Deserialization of Untrusted Data

CVSS: 8.6v4.0HighAnalyzed
AliasesGHSA-c9rc-mg46-23w3PYSEC-2025-75UBUNTU-CVE-2025-8747
Published 11-08-2025Last modified 26-02-2026
Open JSON

Vulnerability summary

CVE-2025-8747 is a high (CVSS 8.6) Deserialization of Untrusted Data vulnerability in google keras. EPSS estimates a 0.12% chance of exploitation in the next 30 days, higher than 1.5% of scored vulnerabilities.

Overall risk
34/100
0100
Medium · default modelHow is this calculated?
CVSS score
8.6/10
010
High · v4.0 (CVE.org)
EPSS score
0.12%
0%100%
Low · 0.00 pp in 30 days
KEVNot listedIn no KEV catalog
RansomwareNone recordedNo known use recorded
Public exploits0No public exploits recorded

Advisory lineage

Upstream 0 · Downstream 0 · Related 1
CGA-GX5F-5Q53-H674

Timeline

  1. 11-08-2025
    PublishedVulnerability first disclosed.
  2. 26-02-2026
    Last modifiedVulnerability information updated.

Description

A safe mode bypass vulnerability in the `Model.load_model` method in Keras versions 3.0.0 through 3.10.0 allows an attacker to achieve arbitrary code execution by convincing a user to load a specially crafted `.keras` model archive.

CVSS metrics

7 scores · 5 sources · 2 versions

EPSS trend

Probability

0.12%

Chance of exploitation in the next 30 days

Percentile

2%

Versus all other CVEs

30-day change

0.00 pp

From 0.12% on 18-09-2026

18-09-202618-10-2026
EPSS score
0.12%

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

Chainguard2 products
  • tensorflow-cpu-jupyter< 2.19.0-r7
  • tensorflow-gpu-jupyter< 2.19.0-r6
Google1 product
  • keras≥ 3.0.0, ≤ 3.10.0
Keras1 product
  • keras≥ 3.0.0, ≤ 3.10.0
PyPI1 product
  • keras≥ 3.0.0, < 3.11.0
Ubuntu1 product
  • kerasAll versions (No fix recorded)

5 vendors · 6 products

References

12 links
Show all 12 references