← Back to search

CVE-2026-119706

9.8 CRITICALpublic exploit available

Published 2026-10-30 · Updated 2026-10-30

AI risk analysis

Summary
InternLM LMDeploy versions 0.9.0 through 0.18.0 contain an unauthenticated remote code execution (RCE) flaw in the /update_weights endpoint due to unsafe deserialization of pickled data via ForkingPickler.loads without validation, allowing attackers to execute arbitrary OS commands as the api_server process.
Exploitability
Exploitation is trivial for remote attackers since no authentication or user interaction is required and the vulnerable endpoint accepts base64-encoded malicious payloads directly.
Blast radius
Successful exploitation grants full system compromise of the host running the api_server, enabling data theft, service disruption, or lateral movement within the infrastructure.
Detection
Defenders should monitor for anomalous POST requests to /update_weights containing base64-encoded payloads with pickle-like signatures (e.g., 'B64:800495...' or 'B64:80027d...'), especially when accompanied by child process creation or unexpected network connections from the api_server process.
Prioritized remediation
Upgrade to LMDeploy version 0.19.0 or later, which addresses this vulnerability by validating and sanitizing serialized data before deserialization.
rcepickle-deserializationunauthenticatedweb-apiics

Analysis generated locally by garnet-coder (no data left the box). AI-assisted — verify against primary sources before acting.

NVD description

InternLM LMDeploy 0.9.0 through 0.18.0 contains an unauthenticated remote code execution vulnerability that allows remote attackers to run OS commands by sending pickled data to the /update_weights endpoint. Attackers can submit a base64-encoded crafted pickle in serialized_named_tensors, which ForkingPickler.loads deserializes without validation, executing commands as the api_server process.

CVSS vector

CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H

Weaknesses

CWE-502

Public exploit & PoC references

All references

Source data: NVD (nvd.nist.gov), public domain. Exploit-DB.ai adds local AI analysis for defensive use only.

Related CVEs

Related by shared AI tags and CWE weakness class. Browse the full archive.