CVE-2026-90777
ESPnet before 202609 deserializes pretrained model checkpoints using torch.load with weights_only=False, allowing arbitrary code execution from attacker-supplied files. Attackers can craft malicious checkpoint files that execute code during deserialization when loaded through the initialization or fine-tuning path.
Scoring
- Severity
- HIGH
- CVSS base score
- 8.8
- CVSS vector
- CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:P/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N
- EPSS probability
- 0.51%
- CWE
- CWE-502
- Published
- 2026-09-13
- Last modified
- 2026-09-14
Affected products
- espnet espnet
- espnet espnet
Weakness type
Related vulnerabilities
- CVE-2026-87719 — Deserialization of Untrusted Data in GitLab
- CVE-2026-72649 — Deserialization of Untrusted Data in Elasticsearch Leading to Remote Code Execution
- CVE-2026-82222 — WordPress GiveWP plugin <= 4.16.7.1 - Remote Code Execution (RCE) vulnerability
- CVE-2026-70416 — Dell ObjectScale, versions prior to 4.4.0.0, contains a Deserialization of Untrusted Data vulnerability. An unauthentica
- CVE-2026-17061 — Deserialization of Untrusted Data Vulnerability in SIMULIA Execution Engine from Release 2023 through Release 2026
- CVE-2026-4703 — WS Form LITE <= 1.10.80 - Unauthenticated PHP Object Injection via Form Submission
- CVE-2026-20307 — Cisco Identity Services Engine Remote Code Execution Vulnerability
- CVE-2026-12650 — A Deserialization of Untrusted Data vulnerability in Ivanti Neurons for ITSM before 2026.2 allows a remote authenticated