Arbitrary Code Execution Vulnerability in SpeechBrain by Lium
CVE-2026-10036

8.7HIGH

Key Information:

Vendor
CVE Published:
27 August 2026

What is CVE-2026-10036?

The SpeechBrain framework, prior to version 1.1.1, is susceptible to an arbitrary code execution vulnerability. This occurs due to improper handling of a crafted CKPT.yaml checkpoint metadata file during candidate enumeration within the Checkpointer.recover_if_possible() function. Attackers can exploit this flaw by embedding harmful Python object construction tags, such as !!python/object/apply, into the CKPT.yaml file located in the configured checkpoint path. This enables them to execute arbitrary code during the candidate discovery phase, regardless of whether the malicious checkpoint is actually chosen for recovery.

Affected Version(s)

speechbrain 0

References

CVSS V4

Score:
8.7
Severity:
HIGH
Confidentiality:
High
Integrity:
High
Availability:
High
Attack Vector:
Network
Attack Complexity:
Low
Attack Required:
None
Privileges Required:
Undefined
User Interaction:
Unknown

Timeline

  • Vulnerability published

  • Vulnerability Reserved

Credit

Sai Teja Erukude
VulnCheck
.