LMDeploy 0.9.0 through 0.18.0 Unauthenticated RCE via /update_weights Pickle Deserialization

Summary

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.

Remediation

1. Inventory every InternLM lmdeploy deployment, including staging.
2. No fixed version appears in the public sources yet. Check the references below and the vendor directly.

References

• Patchhttps://github.com/InternLM/lmdeploy/pull/4964
• Advisoryhttps://www.vulncheck.com/advisories/lmdeploy-0.9.0-through-0.18.0-unauthenticated-rce-via-update-weights-pickle-deserialization
• Referencehttps://github.com/InternLM/lmdeploy
• Referencehttps://github.com/InternLM/lmdeploy/blob/v0.18.0/lmdeploy/serve/openai/endpoints/management.py#L55-L60
• Referencehttps://github.com/InternLM/lmdeploy/blob/v0.18.0/lmdeploy/turbomind/turbomind.py#L345
• Referencehttps://github.com/InternLM/lmdeploy/issues/4698