Arbitrary Code Execution via Crafted Keras Config for Model Loading
41Vexday Risk Score
No sign of exploitation. It has a public proof of concept.
ssvc Attendcvss 7.3epss 2.6%
from disclosure to weapon127 days
Published on NVDMar 11
1st PoC+127d
exploitation probability
2.6%top 16% of all CVEs
observed exploitation
nono source reports it
2 public exploit(s)
The Keras Model.load_model function permits arbitrary code execution, even with safe_mode=True, through a manually constructed, malicious .keras archive. By altering the config.json file within the archive, an attacker can specify arbitrary Python modules and functions, along with their arguments, to be loaded and executed during model loading.
CVSS:4.0/AV:L/AC:L/AT:P/PR:L/UI:A/VC:H/VI:H/VA:H/SC:H/SI:H/SA:H
Affected products
Google · Keraspublic PoCs found — 2
exploitdbwww.exploit-db.com/exploits/52359unverifiedgithubgithub.com/ChCh0i/cve-2025-1550★ 0⚠ Public resources, to assess the exposure of systems you control or are authorized to test. Test only with authorization.